Face_recognition_ResNet.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "91061806",
   "metadata": {},
   "outputs": [],
   "source": [
    "#install dependencies if not installed\n",
    "!pip install opencv-python\n",
    "!pip install dlib -q\n",
    "!wget http://dlib.net/files/shape_predictor_5_face_landmarks.dat.bz2\n",
    "!bzip2 -d shape_predictor_5_face_landmarks.dat.bz2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "84d8e740",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import random\n",
    "from tqdm import tqdm\n",
    "from typing import Union, List\n",
    "\n",
    "import dlib\n",
    "from tqdm import tqdm\n",
    "import numpy as np\n",
    "from imutils import face_utils\n",
    "from PIL import Image\n",
    "from matplotlib import pyplot as plt\n",
    "import matplotlib.patches as patches\n",
    "\n",
    "import transformers\n",
    "import torch\n",
    "from torch import nn\n",
    "import torch.nn.functional as F\n",
    "from torch.utils.data import Dataset, DataLoader, random_split\n",
    "from torchvision import datasets, transforms\n",
    "\n",
    "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "578471a8",
   "metadata": {},
   "source": [
    "Let's start by loading the LFW dataset. We'll create two subsets: a training set (lfw) and a testing set (lfw_test).\n",
    "- \"train\" specifies the training subset.\n",
    "- \"test\" specifies the testing subset.\n",
    "- download=True downloads the dataset if it's not already available locally.\n",
    "\n",
    "Then we will define a function to detect faces, here I use simple dblib package for that. However, it's worth noting that you can also explore other deep learning-based approaches like MTCNN or Single Shot MultiBox Detector (SSD) for face detection."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "e1e82722",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Files already downloaded and verified\n",
      "Files already downloaded and verified\n"
     ]
    }
   ],
   "source": [
    "lfw = datasets.LFWPeople(\"./data\", \"train\", download=True)\n",
    "lfw_test = datasets.LFWPeople(\"./data\", \"test\", download=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "58e00516",
   "metadata": {},
   "outputs": [],
   "source": [
    "def detect_faces(img: Union[np.array, Image.Image], transforms: transforms.Compose=None, size: int=250):\n",
    "    \"\"\"\n",
    "    Detect faces in an input image and return information about the detected faces.\n",
    "\n",
    "    Args:\n",
    "        img (Union[np.array, Image.Image]): Input image containing faces.\n",
    "        transforms (transforms.Compose, optional): A transformation pipeline to apply to detected faces.\n",
    "        size (int, optional): Size of the detected face images. Default is 250 pixels.\n",
    "\n",
    "    Returns:\n",
    "        List[Tuple[np.array, Tuple[int, int, int, int], np.array]]: A list of tuples containing:\n",
    "            - Detected face image as a NumPy array.\n",
    "            - Bounding box coordinates (x, y, width, height) of the detected face.\n",
    "            - Facial landmarks as a NumPy array.\n",
    "\n",
    "    Notes:\n",
    "        - The input image can be either a NumPy array or a PIL Image.\n",
    "        - If a transformation pipeline is provided, it will be applied to the detected faces.\n",
    "        - The function uses the dlib library for face detection and facial landmark prediction.\n",
    "        - Each tuple in the returned list corresponds to a detected face in the input image.\n",
    "    \"\"\"\n",
    "    res = []\n",
    "    detector = dlib.get_frontal_face_detector()\n",
    "    sp = dlib.shape_predictor(\"./shape_predictor_5_face_landmarks.dat\")\n",
    "\n",
    "    if isinstance(img, Image.Image):\n",
    "        img = np.array(img)\n",
    "    detections = detector(img, 1)\n",
    "\n",
    "    for detection in detections:\n",
    "        img_shape = sp(img, detection)\n",
    "        detected_face = dlib.get_face_chip(img, img_shape, size=size)\n",
    "        if transforms:\n",
    "            detected_face = transforms(detected_face)\n",
    "\n",
    "        shape = face_utils.shape_to_np(img_shape)\n",
    "        box_coord = face_utils.rect_to_bb(detection)\n",
    "        res.append((detected_face, box_coord, shape))\n",
    "    return res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "75a2cfc2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "processed, box, shape = detect_faces(lfw[0][0])[0]\n",
    "x, y, w, h = box\n",
    "orig = np.array(lfw[0][0])\n",
    "\n",
    "plt.figure()\n",
    "f, axarr = plt.subplots(1, 3, figsize=(10, 10)) \n",
    "axarr[0].imshow(orig)\n",
    "axarr[1].imshow(processed)\n",
    "axarr[2].imshow(orig)\n",
    "\n",
    "rect = patches.Rectangle((x, y), w, h, linewidth=1, edgecolor='r', facecolor='none')\n",
    "axarr[2].add_patch(rect)\n",
    "for (x, y) in shape:\n",
    "    circle = plt.Circle((x, y), 1, color='r')\n",
    "    axarr[2].add_patch(circle)\n",
    "\n",
    "axarr[0].title.set_text('Original Image')\n",
    "axarr[1].title.set_text('Aligned Image')\n",
    "axarr[2].title.set_text('Detected Face')\n",
    "\n",
    "for ax in axarr:\n",
    "    ax.set_xticks([])\n",
    "    ax.set_yticks([])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "620f0881",
   "metadata": {},
   "outputs": [],
   "source": [
    "# This code segment reads information about the LFW (Labeled Faces in the Wild) dataset from a text file,\n",
    "# organizes it into dictionaries, and preprocesses the images of faces using a face detection function.\n",
    "\n",
    "with open(\"./data/lfw-py/lfw-names.txt\", \"r\") as f:\n",
    "    names = f.readlines()\n",
    "\n",
    "names_dict = {}\n",
    "name2img = {}\n",
    "\n",
    "normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])\n",
    "transform = transforms.Compose([\n",
    "    transforms.ToTensor(),\n",
    "    normalize\n",
    "])\n",
    "\n",
    "for el in tqdm(names):\n",
    "    name, num_images = el.split(\"\\t\")\n",
    "    names_dict[name] = int(num_images.strip())\n",
    "    name2img[name] = []\n",
    "    img_folder = os.path.join(\"./data/lfw-py/lfw_funneled\", name)\n",
    "    all_images = os.listdir(img_folder)\n",
    "    for img in all_images:\n",
    "        img_file = os.path.join(img_folder, img)\n",
    "        image = Image.open(img_file)\n",
    "        \n",
    "        faces = detect_faces(image, transforms=transform)\n",
    "        if faces:\n",
    "            face = faces[0][0]\n",
    "            name2img[name].append(face)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "21973ac0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# I already saved detected faces, so I'll just load them\n",
    "import pickle\n",
    "\n",
    "with open(\"./data/lfw-py/lfw-names.txt\", \"r\") as f:\n",
    "    names = f.readlines()\n",
    "\n",
    "with open(\"./name2img.pickle\", \"rb\") as f:\n",
    "    name2img = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "bfbb5253",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5749/5749 [00:00<00:00, 708915.56it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Abdoulaye_Wade 4 3\n",
      "Angela_Merkel 5 4\n",
      "Anna_Kournikova 12 11\n",
      "Ariel_Sharon 77 76\n",
      "Arlen_Specter 3 2\n",
      "Arnold_Schwarzenegger 42 40\n",
      "Brandon_Webb 1 0\n",
      "Brian_Schneider 1 0\n",
      "Britney_Spears 14 13\n",
      "Chris_Andrews 1 0\n",
      "Colin_Montgomerie 5 4\n",
      "Dale_Earnhardt_Jr 4 3\n",
      "Derrick_Rodgers 1 0\n",
      "Don_Hewitt 1 0\n",
      "Donald_Rumsfeld 121 120\n",
      "Elena_de_Chavez 1 0\n",
      "Elisabeth_Schumacher 2 1\n",
      "Ernie_Stewart 1 0\n",
      "Franz_Muentefering 4 3\n",
      "George_Galloway 4 3\n",
      "George_W_Bush 530 529\n",
      "Guillermo_Coria 30 28\n",
      "Hernan_Diaz 1 0\n",
      "Ira_Einhorn 1 0\n",
      "Jacques_Chirac 52 50\n",
      "James_McGreevey 4 3\n",
      "Jane_Fonda 2 1\n",
      "Jaouad_Gharib 1 0\n",
      "Jeff_Feldman 1 0\n",
      "Jelena_Dokic 8 7\n",
      "Jessica_Lynch 2 1\n",
      "Joe_Vandever 1 0\n",
      "John_Burkett 1 0\n",
      "John_Manley 7 6\n",
      "John_Thune 1 0\n",
      "Jon_Stewart 1 0\n",
      "Kaye_Young 1 0\n",
      "Kultida_Woods 1 0\n",
      "Luis_Horna 6 5\n",
      "Mark_Heller 2 1\n",
      "Meryl_Streep 15 14\n",
      "Nabil_Shaath 3 2\n",
      "Nancy_Pelosi 15 14\n",
      "Rob_Moore 1 0\n",
      "Rob_Ramsay 1 0\n",
      "Robert_Zoellick 7 6\n",
      "Saddam_Hussein 23 22\n",
      "Saeb_Erekat 2 1\n",
      "Tatiana_Gratcheva 1 0\n",
      "Thomas_Birmingham 2 1\n",
      "Tung_Chee-hwa 9 8\n",
      "Vecdi_Gonul 1 0\n",
      "Yasser_Arafat 8 7\n",
      "Yoelbi_Quesada 1 0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "# After detecting faces, some people's images were discarded for not finding the face of the photo\n",
    "# You can use a better face detection like MTCNN\n",
    "\n",
    "names_dict = {}\n",
    "names_without_image = []\n",
    "names_with_less_image = []\n",
    "for el in tqdm(names):\n",
    "    name, num_images = el.split(\"\\t\")\n",
    "    num = int(num_images.strip())\n",
    "    names_dict[name] = num\n",
    "    detected_num = len(name2img[name])\n",
    "    if detected_num == 0:\n",
    "        names_without_image.append(name)\n",
    "    elif num > detected_num:\n",
    "        names_with_less_image.append(name)\n",
    "    if detected_num != num:\n",
    "        print(name, num, detected_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "a16d8bb1",
   "metadata": {},
   "outputs": [],
   "source": [
    "no_face_found = [k for k, v in name2img.items() if not v]\n",
    "names_list = [n.split(\"\\t\")[0] for n in names if n.split(\"\\t\")[0] not in no_face_found]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "35d7a9db",
   "metadata": {},
   "outputs": [],
   "source": [
    "triplets = []\n",
    "for name, images in name2img.items():\n",
    "    if len(images) >= 2:\n",
    "        negative_name = random.choice([n for n in names_list if n!=name])\n",
    "        anchor, pos = random.sample(images, 2)\n",
    "        neg = random.choice(name2img[negative_name])\n",
    "\n",
    "        triplets.append((anchor, pos, neg, name, negative_name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "c8d6b3d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "contrastive_set = []\n",
    "K = 30\n",
    "for name, images in name2img.items():\n",
    "    if len(images) >= 2:\n",
    "        negatives_names = random.sample([n for n in names_list if n!=name], K)\n",
    "        query, pos = random.sample(images, 2)\n",
    "        elements = [pos]\n",
    "        for neg_name in negatives_names:\n",
    "            neg_imgs = name2img[neg_name]\n",
    "            neg = random.choice(neg_imgs)\n",
    "            elements.append(neg)\n",
    "        contrastive_set.append((query, elements, name))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ad486318",
   "metadata": {},
   "source": [
    "In this section, we'll create a custom dataset class called `TripletDataset` tailored for triplet loss-based training.\n",
    "Triplet loss is commonly used in similarity learning tasks, such as face recognition and image retrieval.\n",
    "This dataset class is designed to efficiently handle triplets of data during model training.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "16fe13e2",
   "metadata": {},
   "outputs": [],
   "source": [
    "class TripletDataset(Dataset):\n",
    "    \"\"\"\n",
    "    Custom dataset class for triplet loss-based training.\n",
    "\n",
    "    Args:\n",
    "        triplets (List[Tuple]): A list of triplets, where each triplet contains three samples:\n",
    "            - Anchor: The reference sample.\n",
    "            - Positive: A sample similar to the anchor.\n",
    "            - Negative: A sample dissimilar to the anchor.\n",
    "        transform (callable, optional): A data transformation to be applied to each sample.\n",
    "    \"\"\"\n",
    "    def __init__(self, triplets, transform=None):\n",
    "        self.triplets = triplets\n",
    "        self.transform = transform\n",
    "        \n",
    "    def __getitem__(self, index):\n",
    "        anchor, pos, neg, name_pos, name_neg = self.triplets[index]\n",
    "        if self.transform:\n",
    "            anchor = self.transform(anchor)\n",
    "            pos = self.transform(pos)\n",
    "            neg = self.transform(neg)\n",
    "        return anchor, pos, neg\n",
    "        \n",
    "    def __len__(self):\n",
    "        return len(self.triplets)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "d5964074",
   "metadata": {},
   "outputs": [],
   "source": [
    "class ContrastiveDataset(Dataset):\n",
    "    \"\"\"\n",
    "    Custom dataset class for contrastive loss-based training.\n",
    "\n",
    "    Args:\n",
    "        contrastive_set (List[Tuple]): A list of tuples, where each tuple contains:\n",
    "            - Reference Sample: A reference sample.\n",
    "            - Negative Samples: A list of samples where the first one is similar to the reference \n",
    "                and others are dissimilar to the reference.\n",
    "        transform (callable, optional): A data transformation to be applied to each sample.\n",
    "    \"\"\"\n",
    "    def __init__(self, contrastive_set, transform=None):\n",
    "        self.contrastive_set = contrastive_set\n",
    "        self.transform = transform\n",
    "        \n",
    "    def __getitem__(self, index):\n",
    "        query, elements, name_pos = self.contrastive_set[index]\n",
    "        if self.transform:\n",
    "            query = self.transform(query)\n",
    "            elements = torch.stack([self.transform(el) for el in elements], dim=0)\n",
    "        else:\n",
    "            elements = torch.stack([el for el in elements], dim=0)\n",
    "        return torch.cat((query.unsqueeze(0), elements))\n",
    "        \n",
    "    def __len__(self):\n",
    "        return len(self.contrastive_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "84ec5174",
   "metadata": {},
   "outputs": [],
   "source": [
    "transform_aug = transforms.Compose([\n",
    "    transforms.RandomHorizontalFlip(),\n",
    "    transforms.RandomRotation(degrees=5),\n",
    "])\n",
    "\n",
    "train_dataset = TripletDataset(triplets[:1500], transform=transform_aug)\n",
    "test_dataset = TripletDataset(triplets[1500:])\n",
    "train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "b55689e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset_cont = ContrastiveDataset(contrastive_set[:1500], transform=transform_aug)\n",
    "test_dataset_cont = ContrastiveDataset(contrastive_set[1500:])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "906140cd",
   "metadata": {},
   "source": [
    "### Start Defining Loss Functions and Model Architecture\n",
    "\n",
    "Now that we have preprocessed our dataset, it's time to set up the loss functions and define our model architecture. In this section, we'll introduce two common loss functions used in similarity learning tasks: Triplet Loss and Contrastive Loss.\n",
    "\n",
    "#### Triplet Loss\n",
    "\n",
    "Triplet Loss is a popular loss function for training models in tasks where we want similar samples to be closer in feature space and dissimilar samples to be farther apart.\n",
    "\n",
    "#### Contrastive Loss \n",
    "It is another loss function used for similarity learning. It encourages positive samples to be close and negative samples to be far apart in the feature space."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "f57184a9",
   "metadata": {},
   "outputs": [],
   "source": [
    "from typing import Union, Tuple, List, Iterable, Dict\n",
    "import torch.nn.functional as F\n",
    "from enum import Enum\n",
    "\n",
    "\n",
    "class DistanceMetrics(Enum):\n",
    "    COSINE = lambda x, y: 1 - F.cosine_similarity(x, y)\n",
    "    EUCLIDEAN = lambda x, y: F.pairwise_distance(x, y, p=2)\n",
    "    MANHATTAN = lambda x, y: F.pairwise_distance(x, y, p=1)\n",
    "\n",
    "\n",
    "class TripletLoss(nn.Module):\n",
    "    def __init__(self, distance_metric=DistanceMetrics.EUCLIDEAN, triplet_margin: float=2.0):\n",
    "        super(TripletLoss, self).__init__()\n",
    "        self.distance_metric = distance_metric\n",
    "        self.triplet_margin = triplet_margin\n",
    "\n",
    "    def forward(self, anchor_feats: torch.Tensor, pos_feats: torch.Tensor, neg_feats: torch.Tensor):\n",
    "        distance_pos = self.distance_metric(anchor_feats, pos_feats)\n",
    "        distance_neg = self.distance_metric(anchor_feats, neg_feats)\n",
    "\n",
    "        losses = F.relu(distance_pos - distance_neg + self.triplet_margin)\n",
    "        return losses.mean()\n",
    "\n",
    "\n",
    "class TripletLossHardNegatives(nn.Module):\n",
    "    def __init__(self, distance_metric=DistanceMetrics.EUCLIDEAN, triplet_margin: float=0.2):\n",
    "        super(TripletLossHardNegatives, self).__init__()\n",
    "        self.distance_metric = distance_metric\n",
    "        self.margin = triplet_margin\n",
    "\n",
    "    def compute_distance_matrix(self, anchor, positive, negative):\n",
    "        \"\"\"\n",
    "        Compute distance matrix between anchor, positive, and negative samples.\n",
    "        \"\"\"\n",
    "        distance_matrix = torch.zeros(anchor.size(0), 3)\n",
    "        distance_matrix[:, 0] = self.distance_metric(anchor, anchor)\n",
    "        distance_matrix[:, 1] = self.distance_metric(anchor, positive)\n",
    "        distance_matrix[:, 2] = self.distance_metric(anchor, negative)\n",
    "        return distance_matrix\n",
    "\n",
    "    def forward(self, anchor, positive, negative):\n",
    "        \"\"\"\n",
    "        Compute triplet loss using the batch hard strategy.\n",
    "        \"\"\"\n",
    "        distance_matrix = self.compute_distance_matrix(anchor, positive, negative)\n",
    "        hard_negative = torch.argmax(distance_matrix[:, 2])\n",
    "        loss = torch.max(torch.tensor(0.0), distance_matrix[:, 0] - distance_matrix[:, 1] + self.margin)\n",
    "        loss += torch.max(torch.tensor(0.0), distance_matrix[:, 0][hard_negative] - distance_matrix[:, 2] + self.margin)\n",
    "        return torch.mean(loss)\n",
    "\n",
    "\n",
    "class ContrastiveLoss(nn.Module):\n",
    "    def __init__(self, in_batch_only: bool=True):\n",
    "        super(ContrastiveLoss, self).__init__()\n",
    "        self.in_batch_only = in_batch_only\n",
    "\n",
    "    def forward(self, features: torch.Tensor):\n",
    "        batch_size, num_samples, dim = features.size()\n",
    "\n",
    "        query, elements = features[:, 0, :], features[:, 1:, :]\n",
    "        elements = elements.reshape(-1, dim)\n",
    "        scores = torch.matmul(query, elements.transpose(0, 1))\n",
    "        # because the positive class is always the first one\n",
    "        labels = torch.arange(0, batch_size * (num_samples-1), num_samples-1).to(scores.device)\n",
    "    \n",
    "        if self.in_batch_only:\n",
    "            scores_per_batch = []\n",
    "            for batch, lbl in enumerate(labels):\n",
    "                scores_curr_batch = scores[batch, lbl:lbl+(num_samples-1)]\n",
    "                scores_per_batch.append(scores_curr_batch)\n",
    "            new_scores = torch.stack(scores_per_batch, dim=0)\n",
    "            new_labels = torch.zeros(batch_size).long().to(scores.device)\n",
    "            loss = F.cross_entropy(new_scores, new_labels)\n",
    "            max_score, max_idxs = torch.max(new_scores, -1)\n",
    "            correct_predictions_count = (max_idxs == new_labels).sum().cpu().numpy()\n",
    "        else:            \n",
    "            loss = F.cross_entropy(scores, labels)\n",
    "            softmax_scores = F.log_softmax(scores, dim=1)\n",
    "            max_score, max_idxs = torch.max(softmax_scores, -1)\n",
    "            correct_predictions_count = (max_idxs == labels).sum().cpu().numpy()\n",
    "        return loss, correct_predictions_count"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d24dcdfc",
   "metadata": {},
   "source": [
    "For our image recognition task, we'll use a simplified version of the ResNet architecture. ResNet (Residual Networks) is a groundbreaking deep neural network architecture designed to tackle the challenges of training very deep networks. It achieved this by introducing skip connections or residual blocks.\n",
    "\n",
    "Our ResNet architecture consists of multiple BlocksGroups, where each group contains a set of ResNet blocks. The `ResNet` class defines this architecture.\n",
    "\n",
    "\n",
    "#### Key Layers in ResNet\n",
    "\n",
    "- **Convolutional Layers**: These layers perform feature extraction from raw data.\n",
    "\n",
    "- **Batch Normalization**: It normalizes layer outputs, aiding training stability.\n",
    "\n",
    "- **Activation Function (ReLU)**: Adds non-linearity for feature representation.\n",
    "\n",
    "- **Max Pooling**: Reduces spatial dimensions, preserving important information.\n",
    "\n",
    "- **Adaptive Average Pooling**: Creates fixed-size feature maps for global pooling.\n",
    "\n",
    "- **ResNet Blocks**: Contain skip connections that ease training of extremely deep networks.\n",
    "\n",
    "#### Why ResNet?\n",
    "\n",
    "ResNet's innovation in residual blocks enables training of exceptionally deep neural networks. Skip connections mitigate vanishing gradient issues and allow for efficient feature learning. ResNet excels in computer vision tasks like image classification, object detection, and segmentation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "36656eb7",
   "metadata": {},
   "outputs": [],
   "source": [
    "class ResNetBlock(nn.Module):\n",
    "    def __init__(self, in_channels: int, out_channels: int, downsample: nn.Module=None, stride: int=1):\n",
    "        super().__init__()\n",
    "        self.conv1 = nn.Conv2d(in_channels, in_channels, 1, stride=stride)\n",
    "        self.conv2 = nn.Conv2d(in_channels, out_channels, 3, padding=1, stride=stride)\n",
    "        self.conv3 = nn.Conv2d(out_channels, out_channels, 1)\n",
    "        self.batch_norm1 = nn.BatchNorm2d(in_channels)\n",
    "        self.batch_norm2 = nn.BatchNorm2d(out_channels)\n",
    "        self.batch_norm3 = nn.BatchNorm2d(out_channels)\n",
    "        self.activation = nn.ReLU()\n",
    "        self.downsample = downsample or nn.Identity()\n",
    "\n",
    "    def forward(self, x):\n",
    "        res = self.downsample(x)\n",
    "\n",
    "        x = self.conv1(x)\n",
    "        x = self.batch_norm1(x)\n",
    "        x = self.activation(x)\n",
    "        x = self.conv2(x)\n",
    "        x = self.batch_norm2(x)\n",
    "        x = self.conv3(x)\n",
    "        x = self.batch_norm3(x)\n",
    "        x += res\n",
    "        x = self.activation(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "3e88f735",
   "metadata": {},
   "outputs": [],
   "source": [
    "class BlocksGroup(nn.Module):\n",
    "    def __init__(self, in_channels, out_channels: int, blocks: int, stride: int=1):\n",
    "        super().__init__()\n",
    "        downsample = nn.Identity()\n",
    "        if stride != 1 or in_channels != out_channels:\n",
    "            downsample = nn.Sequential(\n",
    "                nn.Conv2d(in_channels, out_channels, stride),\n",
    "                nn.BatchNorm2d(out_channels),\n",
    "            )\n",
    "\n",
    "        layers = []\n",
    "        layers.append(ResNetBlock(in_channels, out_channels, downsample, stride=stride))\n",
    "        in_channels = out_channels\n",
    "        for _ in range(1, blocks):\n",
    "            layers.append(ResNetBlock(in_channels, out_channels))\n",
    "        self.blocks = nn.Sequential(*layers)\n",
    "\n",
    "    def forward(self, x):\n",
    "        return self.blocks(x)\n",
    "\n",
    "\n",
    "class ResNet(nn.Module):\n",
    "    def __init__(self, blocks, channels, num_classes):\n",
    "        super().__init__()\n",
    "        \n",
    "        self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3)\n",
    "        self.batch_norm = nn.BatchNorm2d(64)\n",
    "        self.activation = nn.ReLU(inplace=True)\n",
    "        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)\n",
    "        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))\n",
    "\n",
    "        init_channel = 64\n",
    "        self.block_layers = nn.Sequential()\n",
    "        for block, out_channel in zip(blocks, channels):\n",
    "            block_layer = BlocksGroup(init_channel, out_channel, block)\n",
    "            self.block_layers.append(block_layer)\n",
    "            init_channel = out_channel\n",
    "        self.classifier = nn.Sequential(\n",
    "            nn.Flatten(),\n",
    "            nn.Linear(512, num_classes),\n",
    "        )\n",
    "      \n",
    "    def forward(self, x):\n",
    "        x = self.batch_norm(self.conv1(x))\n",
    "        x = self.maxpool(self.activation(x))\n",
    "        x = self.block_layers(x)\n",
    "        x = self.avgpool(x)\n",
    "        x = torch.flatten(x, 1)\n",
    "        x = self.classifier(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4fbe6a81",
   "metadata": {},
   "source": [
    "With our ResNet architecture defined, it's time to create an instance of the model, set up the optimizer, loss function, and configure the training process.\n",
    "\n",
    "We're using a ResNet model, specifying the number of blocks in each stage (ResNet [2, 3, 3, 4]), the number of channels in each stage, and the number of output classes (512 in this case). Adjust the architecture (number of blocks, channels, etc.) based on your dataset's characteristics and available computing resources.\n",
    "\n",
    "We select the optimizer that will update the model's parameters during training. AdamW is a variant of the Adam optimizer that helps stabilize training. The loss function quantifies the difference between predicted and actual values. Here, we're using Triplet Loss, suitable for similarity learning tasks.\n",
    "\n",
    "The hyperparameters control the training process. `grad_accum_steps` determines how often gradients are accumulated before a backward pass, while `num_train_epochs` sets the number of training epochs. A learning rate scheduler adjusts the learning rate during training. Here, we use a linear scheduler with warm-up.\n",
    "\n",
    "\n",
    "**Tips**:\n",
    "- Experiment with different optimizers and learning rates to find the best combination for your specific task.\n",
    "- Learning rate (lr) is a crucial hyperparameter. Start with a reasonable value (e.g., 1e-3) and adjust as needed.\n",
    "- Gradual accumulation of gradients (`grad_accum_steps`) can help reduce memory usage during training, especially on GPUs with limited memory.\n",
    "- Monitor training progress and validation performance to decide the number of epochs (`num_train_epochs`). Early stopping may be useful.\n",
    "- Learning rate scheduling can help fine-tune training. Experiment with different schedules (e.g., step decay, cosine annealing) to find the best one.\n",
    "- Warm-up steps allow the model to start with a smaller learning rate and gradually increase it. This can lead to better convergence.\n",
    "- Utilize multiple GPUs if you have access to them to speed up training.\n",
    "- Ensure your model and data are correctly placed on the chosen device for training."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fb3e327c",
   "metadata": {},
   "source": [
    "## Triplet Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "41cb246f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DataParallel(\n",
       "  (module): ResNet(\n",
       "    (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3))\n",
       "    (batch_norm): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    (activation): ReLU(inplace=True)\n",
       "    (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
       "    (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
       "    (block_layers): Sequential(\n",
       "      (0): BlocksGroup(\n",
       "        (blocks): Sequential(\n",
       "          (0): ResNetBlock(\n",
       "            (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "          (1): ResNetBlock(\n",
       "            (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (1): BlocksGroup(\n",
       "        (blocks): Sequential(\n",
       "          (0): ResNetBlock(\n",
       "            (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Sequential(\n",
       "              (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "              (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            )\n",
       "          )\n",
       "          (1): ResNetBlock(\n",
       "            (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "          (2): ResNetBlock(\n",
       "            (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (2): BlocksGroup(\n",
       "        (blocks): Sequential(\n",
       "          (0): ResNetBlock(\n",
       "            (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(128, 264, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Sequential(\n",
       "              (0): Conv2d(128, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "              (1): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            )\n",
       "          )\n",
       "          (1): ResNetBlock(\n",
       "            (conv1): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(264, 264, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "          (2): ResNetBlock(\n",
       "            (conv1): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(264, 264, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (3): BlocksGroup(\n",
       "        (blocks): Sequential(\n",
       "          (0): ResNetBlock(\n",
       "            (conv1): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(264, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Sequential(\n",
       "              (0): Conv2d(264, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "              (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            )\n",
       "          )\n",
       "          (1): ResNetBlock(\n",
       "            (conv1): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "          (2): ResNetBlock(\n",
       "            (conv1): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (classifier): Sequential(\n",
       "      (0): Flatten(start_dim=1, end_dim=-1)\n",
       "      (1): Linear(in_features=512, out_features=512, bias=True)\n",
       "    )\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = ResNet([2, 3, 3, 3], [64, 128, 264, 512], 512)\n",
    "optimizer = torch.optim.AdamW(model.parameters(), lr=1e-04)\n",
    "loss_fn = TripletLoss()\n",
    "\n",
    "grad_accum_steps = 8\n",
    "num_train_epochs = 40\n",
    "total_iters = len(train_loader) // grad_accum_steps * num_train_epochs\n",
    "scheduler = transformers.get_linear_schedule_with_warmup(\n",
    "        optimizer, num_warmup_steps=50, num_training_steps=total_iters\n",
    ")\n",
    "\n",
    "# These lines configure model parallelism (use multiple GPUs if available) and specify the target device (e.g., CPU or GPU).\n",
    "if torch.cuda.device_count() > 1:\n",
    "    model = nn.DataParallel(model)\n",
    "    model_accessor = model.module\n",
    "else:\n",
    "    model_accessor = model\n",
    "model.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "03012301",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# used for test loop\n",
    "test_loader = DataLoader(test_dataset, batch_size=32, num_workers=4)\n",
    "margin = 0.2\n",
    "distance = DistanceMetrics.EUCLIDEAN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "8f591156",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 1/40: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:44<00:00,  2.22s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.978\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:04<00:00,  1.33it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 50.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 2/40: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:41<00:00,  2.17s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.759\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:04<00:00,  1.47it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 56.322\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 3/40: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:20<00:00,  1.71s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.591\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:04<00:00,  1.42it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 60.345\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 4/40: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:22<00:00,  1.76s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.533\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:04<00:00,  1.36it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 58.621\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 5/40: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:19<00:00,  1.69s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.41\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:04<00:00,  1.46it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 60.345\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 6/40: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:22<00:00,  1.75s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.326\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.62it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 61.494\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 7/40: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:19<00:00,  1.69s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.297\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.66it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 58.621\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 8/40: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:20<00:00,  1.71s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.22\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.66it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 60.92\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 9/40: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:17<00:00,  1.66s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.111\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.66it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 59.77\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 10/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:18<00:00,  1.66s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.056\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.57it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 62.644\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 11/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:21<00:00,  1.73s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.99\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.50it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 67.816\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 12/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:18<00:00,  1.68s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 1.001\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.65it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 66.092\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 13/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:23<00:00,  1.78s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.92\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:04<00:00,  1.48it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 62.644\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 14/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:19<00:00,  1.69s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.916\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.63it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 66.667\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 15/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:20<00:00,  1.71s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.852\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:04<00:00,  1.30it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 68.966\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 16/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:18<00:00,  1.68s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.826\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.66it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 64.943\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 17/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:19<00:00,  1.68s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.761\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.68it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 64.943\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 18/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:19<00:00,  1.70s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.729\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.67it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 66.667\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 19/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:21<00:00,  1.73s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.717\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:04<00:00,  1.43it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 65.517\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 20/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:21<00:00,  1.74s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.678\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.57it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 69.54\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 21/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:19<00:00,  1.69s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.629\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.59it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 66.667\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 22/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:19<00:00,  1.68s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.6\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.66it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 69.54\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 23/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:17<00:00,  1.66s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.521\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.68it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 65.517\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 24/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:18<00:00,  1.67s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.591\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.71it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 67.816\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 25/40: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [01:19<00:00,  1.69s/batch]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loss: 0.508\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:03<00:00,  1.62it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy test: 67.816\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Epoch 26/40:   2%|██▋                                                                                                                           | 1/47 [00:03<02:50,  3.70s/batch]\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[25], line 15\u001b[0m\n\u001b[1;32m     13\u001b[0m loss \u001b[38;5;241m=\u001b[39m loss_fn(anchor_feats, pos_feats, neg_feats) \u001b[38;5;241m/\u001b[39m grad_accum_steps\n\u001b[1;32m     14\u001b[0m loss\u001b[38;5;241m.\u001b[39mbackward()\n\u001b[0;32m---> 15\u001b[0m avg_loss \u001b[38;5;241m+\u001b[39m\u001b[38;5;241m=\u001b[39m \u001b[43mloss\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mitem\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m     16\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m (i \u001b[38;5;241m+\u001b[39m \u001b[38;5;241m1\u001b[39m) \u001b[38;5;241m%\u001b[39m grad_accum_steps \u001b[38;5;241m==\u001b[39m \u001b[38;5;241m0\u001b[39m:\n\u001b[1;32m     17\u001b[0m     optimizer\u001b[38;5;241m.\u001b[39mstep()\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "# the core training loop iterates through the dataset for num_train_epochs\n",
    "for epoch in range(num_train_epochs):\n",
    "    avg_loss = 0\n",
    "    model_accessor.train()\n",
    "    for i, element in tqdm(enumerate(train_loader), desc=f\"Epoch {epoch+1}/{num_train_epochs}\", \n",
    "                            unit=\"batch\", total=len(train_loader)):\n",
    "        anchor, pos, neg = element\n",
    "        batch_size = anchor.size()[0]\n",
    "        all_inputs = torch.cat((anchor, pos, neg))\n",
    "        all_inputs = all_inputs.to(device)\n",
    "        out = model(all_inputs)\n",
    "        anchor_feats, pos_feats, neg_feats = torch.split(out, batch_size, 0)\n",
    "        loss = loss_fn(anchor_feats, pos_feats, neg_feats) / grad_accum_steps\n",
    "        loss.backward()\n",
    "        avg_loss += loss.item()\n",
    "        if (i + 1) % grad_accum_steps == 0:\n",
    "            optimizer.step()\n",
    "            optimizer.zero_grad()\n",
    "            scheduler.step()\n",
    "\n",
    "    print(f\"Loss: {round(avg_loss/len(train_loader)*grad_accum_steps, 3)}\")\n",
    "    \n",
    "    correct = total = 0\n",
    "    with torch.no_grad():\n",
    "        model_accessor.eval()\n",
    "        for elements in tqdm(test_loader):\n",
    "            anchor, pos, neg = elements\n",
    "            batch_size = anchor.size()[0]\n",
    "            all_inputs = torch.cat((anchor, pos, neg))\n",
    "            all_inputs = all_inputs.to(device)\n",
    "            out = model(all_inputs)\n",
    "            anchor_feats, pos_feats, neg_feats = torch.split(out, batch_size, 0)\n",
    "\n",
    "            pos_dists = distance(anchor_feats, pos_feats)\n",
    "            neg_dists = distance(anchor_feats, neg_feats)\n",
    "\n",
    "            # Calculate accuracy by counting times when the difference between negative and positive images is more than a margin\n",
    "            # and when the positive is closer to anchor than the negative\n",
    "            first_condition = (neg_dists - pos_dists > margin).cpu().numpy().flatten()\n",
    "            second_condition = (pos_dists < neg_dists).cpu().numpy().flatten()\n",
    "            both_conds = (np.logical_and(first_condition, second_condition))\n",
    "            correct += np.sum(both_conds)\n",
    "            total += batch_size\n",
    "\n",
    "    print(f'Accuracy test: {round(correct/total*100, 3)}') "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ac495315",
   "metadata": {},
   "source": [
    "## Contrastive Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "9a1e8c2d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Adjust the batch size since each batch in constrastive learning contains (2 + neagtive_samples_num) elements\n",
    "train_loader_cont = DataLoader(train_dataset_cont, batch_size=2, shuffle=True, num_workers=4)\n",
    "test_loader_cont = DataLoader(test_dataset_cont, batch_size=4, num_workers=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "a7d0498d",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DataParallel(\n",
       "  (module): ResNet(\n",
       "    (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3))\n",
       "    (batch_norm): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "    (activation): ReLU(inplace=True)\n",
       "    (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)\n",
       "    (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))\n",
       "    (block_layers): Sequential(\n",
       "      (0): BlocksGroup(\n",
       "        (blocks): Sequential(\n",
       "          (0): ResNetBlock(\n",
       "            (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "          (1): ResNetBlock(\n",
       "            (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (1): BlocksGroup(\n",
       "        (blocks): Sequential(\n",
       "          (0): ResNetBlock(\n",
       "            (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Sequential(\n",
       "              (0): Conv2d(64, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "              (1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            )\n",
       "          )\n",
       "          (1): ResNetBlock(\n",
       "            (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "          (2): ResNetBlock(\n",
       "            (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (2): BlocksGroup(\n",
       "        (blocks): Sequential(\n",
       "          (0): ResNetBlock(\n",
       "            (conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(128, 264, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Sequential(\n",
       "              (0): Conv2d(128, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "              (1): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            )\n",
       "          )\n",
       "          (1): ResNetBlock(\n",
       "            (conv1): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(264, 264, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "          (2): ResNetBlock(\n",
       "            (conv1): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(264, 264, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "      (3): BlocksGroup(\n",
       "        (blocks): Sequential(\n",
       "          (0): ResNetBlock(\n",
       "            (conv1): Conv2d(264, 264, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(264, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(264, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Sequential(\n",
       "              (0): Conv2d(264, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "              (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            )\n",
       "          )\n",
       "          (1): ResNetBlock(\n",
       "            (conv1): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "          (2): ResNetBlock(\n",
       "            (conv1): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "          (3): ResNetBlock(\n",
       "            (conv1): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "            (conv3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
       "            (batch_norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (batch_norm3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "            (activation): ReLU()\n",
       "            (downsample): Identity()\n",
       "          )\n",
       "        )\n",
       "      )\n",
       "    )\n",
       "    (classifier): Sequential(\n",
       "      (0): Flatten(start_dim=1, end_dim=-1)\n",
       "      (1): Linear(in_features=512, out_features=512, bias=True)\n",
       "    )\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = ResNet([2, 3, 3, 4], [64, 128, 264, 512], 512)\n",
    "optimizer = torch.optim.AdamW(model.parameters(), lr=1e-05)\n",
    "loss_fn = ContrastiveLoss()\n",
    "\n",
    "grad_accum_steps = 16\n",
    "num_train_epochs = 30\n",
    "total_iters = len(train_loader_cont) // grad_accum_steps * num_train_epochs\n",
    "scheduler = transformers.get_linear_schedule_with_warmup(\n",
    "        optimizer, num_warmup_steps=50, num_training_steps=total_iters\n",
    ")\n",
    "\n",
    "# These lines configure model parallelism (use multiple GPUs if available) and specify the target device (e.g., CPU or GPU).\n",
    "if torch.cuda.device_count() > 1:\n",
    "    model = nn.DataParallel(model)\n",
    "    model_accessor = model.module\n",
    "else:\n",
    "    model_accessor = model\n",
    "\n",
    "ckpt = torch.load(\"./lwf_ckpt.pth.tar\")\n",
    "model.load_state_dict(ckpt['model_state_dict'])\n",
    "model.to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fdff53f2",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "for epoch in range(num_train_epochs):\n",
    "    avg_loss = 0\n",
    "    total = corrects = 0\n",
    "    correct_test = total_test = 0\n",
    "    losses = {\"train\": 0, \"test\": 0}\n",
    "    model_accessor.train()\n",
    "    for i, element in tqdm(enumerate(train_loader_cont), desc=f\"Epoch {epoch+1}/{num_train_epochs}\", \n",
    "                            unit=\"batch\", total=len(train_loader_cont)):\n",
    "        bs, num_samples, C, H, W = element.size()\n",
    "        all_inputs = element.view(-1, C, H, W)\n",
    "        all_inputs = all_inputs.to(device)\n",
    "        out = model(all_inputs)\n",
    "        out = out.view(bs, num_samples, -1)\n",
    "        loss, correct = loss_fn(out)\n",
    "        loss /= grad_accum_steps\n",
    "        loss.backward()\n",
    "        losses['train'] += loss.item()\n",
    "        corrects += correct\n",
    "        total += bs\n",
    "        if (i + 1) % grad_accum_steps == 0:\n",
    "            optimizer.step()\n",
    "            optimizer.zero_grad()\n",
    "            scheduler.step()\n",
    "\n",
    "    print(f\"Loss: {round(losses['train']/len(train_loader_cont)*grad_accum_steps, 3)}\")\n",
    "    print(f\"Accuracy: {round(corrects/total*100, 3)}\")\n",
    "    \n",
    "    with torch.no_grad():\n",
    "        model_accessor.eval()\n",
    "        for elements in tqdm(test_loader_cont):\n",
    "            bs, num_samples, C, H, W = elements.size()\n",
    "            all_inputs = elements.view(-1, C, H, W)\n",
    "            all_inputs = all_inputs.to(device)\n",
    "            out = model(all_inputs)\n",
    "            out = out.view(bs, num_samples, -1)\n",
    "            loss, cor = loss_fn(out)\n",
    "            losses['test'] += loss.item()\n",
    "\n",
    "            correct_test += cor\n",
    "            total_test += bs\n",
    "\n",
    "    print(f\"Loss test: {round(losses['test']/len(test_loader_cont), 3)}\")\n",
    "    print(f'Accuracy test: {round(correct_test/total_test*100, 3)}')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "66b03198",
   "metadata": {},
   "source": [
    "## Save the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cb070a31",
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save({\n",
    "        'model_state_dict': model.state_dict(),\n",
    "        'optimizer_state_dict': optimizer.state_dict(),\n",
    "        }, \"./lwf_ckpt.pth.tar\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2102ae8b",
   "metadata": {},
   "source": [
    "## Evaluate with lfw test protocal using pairs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "40a3de0c",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import KFold\n",
    "from scipy import interpolate\n",
    "\n",
    "# LFW functions taken from David Sandberg's FaceNet implementation\n",
    "def distance(embeddings1, embeddings2, distance_metric=0):\n",
    "    if distance_metric==0:\n",
    "        # Euclidian distance\n",
    "        diff = np.subtract(embeddings1, embeddings2)\n",
    "        dist = np.sum(np.square(diff),1)\n",
    "    elif distance_metric==1:\n",
    "        # Distance based on cosine similarity\n",
    "        dot = np.sum(np.multiply(embeddings1, embeddings2), axis=1)\n",
    "        norm = np.linalg.norm(embeddings1, axis=1) * np.linalg.norm(embeddings2, axis=1)\n",
    "        similarity = dot / norm\n",
    "        dist = np.arccos(similarity) / math.pi\n",
    "    else:\n",
    "        raise 'Undefined distance metric %d' % distance_metric\n",
    "\n",
    "    return dist\n",
    "\n",
    "\n",
    "def calculate_roc(thresholds, embeddings1, embeddings2, actual_issame, nrof_folds=10, distance_metric=0, subtract_mean=False):\n",
    "    assert(embeddings1.shape[0] == embeddings2.shape[0])\n",
    "    assert(embeddings1.shape[1] == embeddings2.shape[1])\n",
    "    nrof_pairs = min(len(actual_issame), embeddings1.shape[0])\n",
    "    nrof_thresholds = len(thresholds)\n",
    "    k_fold = KFold(n_splits=nrof_folds, shuffle=False)\n",
    "\n",
    "    tprs = np.zeros((nrof_folds,nrof_thresholds))\n",
    "    fprs = np.zeros((nrof_folds,nrof_thresholds))\n",
    "    accuracy = np.zeros((nrof_folds))\n",
    "\n",
    "    is_false_positive = []\n",
    "    is_false_negative = []\n",
    "\n",
    "    indices = np.arange(nrof_pairs)\n",
    "\n",
    "    for fold_idx, (train_set, test_set) in enumerate(k_fold.split(indices)):\n",
    "        if subtract_mean:\n",
    "            mean = np.mean(np.concatenate([embeddings1[train_set], embeddings2[train_set]]), axis=0)\n",
    "        else:\n",
    "            mean = 0.0\n",
    "        dist = distance(embeddings1-mean, embeddings2-mean, distance_metric)\n",
    "\n",
    "        # Find the best threshold for the fold\n",
    "        acc_train = np.zeros((nrof_thresholds))\n",
    "        for threshold_idx, threshold in enumerate(thresholds):\n",
    "            _, _, acc_train[threshold_idx], _ ,_ = calculate_accuracy(threshold, dist[train_set], actual_issame[train_set])\n",
    "        best_threshold_index = np.argmax(acc_train)\n",
    "        for threshold_idx, threshold in enumerate(thresholds):\n",
    "            tprs[fold_idx,threshold_idx], fprs[fold_idx,threshold_idx], _, _, _ = calculate_accuracy(threshold, dist[test_set], actual_issame[test_set])\n",
    "        _, _, accuracy[fold_idx], is_fp, is_fn = calculate_accuracy(thresholds[best_threshold_index], dist[test_set], actual_issame[test_set])\n",
    "\n",
    "        tpr = np.mean(tprs,0)\n",
    "        fpr = np.mean(fprs,0)\n",
    "        is_false_positive.extend(is_fp)\n",
    "        is_false_negative.extend(is_fn)\n",
    "\n",
    "    return tpr, fpr, accuracy, is_false_positive, is_false_negative\n",
    "\n",
    "\n",
    "def calculate_accuracy(threshold, dist, actual_issame):\n",
    "    predict_issame = np.less(dist, threshold)\n",
    "    tp = np.sum(np.logical_and(predict_issame, actual_issame))\n",
    "    fp = np.sum(np.logical_and(predict_issame, np.logical_not(actual_issame)))\n",
    "    tn = np.sum(np.logical_and(np.logical_not(predict_issame), np.logical_not(actual_issame)))\n",
    "    fn = np.sum(np.logical_and(np.logical_not(predict_issame), actual_issame))\n",
    "\n",
    "    is_fp = np.logical_and(predict_issame, np.logical_not(actual_issame))\n",
    "    is_fn = np.logical_and(np.logical_not(predict_issame), actual_issame)\n",
    "\n",
    "    tpr = 0 if (tp+fn==0) else float(tp) / float(tp+fn)\n",
    "    fpr = 0 if (fp+tn==0) else float(fp) / float(fp+tn)\n",
    "    acc = float(tp+tn)/dist.size\n",
    "    return tpr, fpr, acc, is_fp, is_fn\n",
    "\n",
    "\n",
    "def calculate_val(thresholds, embeddings1, embeddings2, actual_issame, far_target, nrof_folds=10, distance_metric=0, subtract_mean=False):\n",
    "    assert(embeddings1.shape[0] == embeddings2.shape[0])\n",
    "    assert(embeddings1.shape[1] == embeddings2.shape[1])\n",
    "    nrof_pairs = min(len(actual_issame), embeddings1.shape[0])\n",
    "    nrof_thresholds = len(thresholds)\n",
    "    k_fold = KFold(n_splits=nrof_folds, shuffle=False)\n",
    "\n",
    "    val = np.zeros(nrof_folds)\n",
    "    far = np.zeros(nrof_folds)\n",
    "\n",
    "    indices = np.arange(nrof_pairs)\n",
    "\n",
    "    for fold_idx, (train_set, test_set) in enumerate(k_fold.split(indices)):\n",
    "        if subtract_mean:\n",
    "            mean = np.mean(np.concatenate([embeddings1[train_set], embeddings2[train_set]]), axis=0)\n",
    "        else:\n",
    "            mean = 0.0\n",
    "        dist = distance(embeddings1-mean, embeddings2-mean, distance_metric)\n",
    "\n",
    "        # Find the threshold that gives FAR = far_target\n",
    "        far_train = np.zeros(nrof_thresholds)\n",
    "        for threshold_idx, threshold in enumerate(thresholds):\n",
    "            _, far_train[threshold_idx] = calculate_val_far(threshold, dist[train_set], actual_issame[train_set])\n",
    "        if np.max(far_train) >= far_target:\n",
    "            f = interpolate.interp1d(far_train, thresholds, kind='linear')\n",
    "            threshold = f(far_target)\n",
    "        else:\n",
    "            threshold = 0.0\n",
    "\n",
    "        val[fold_idx], far[fold_idx] = calculate_val_far(threshold, dist[test_set], actual_issame[test_set])\n",
    "\n",
    "    val_mean = np.mean(val)\n",
    "    far_mean = np.mean(far)\n",
    "    val_std = np.std(val)\n",
    "    return val_mean, val_std, far_mean\n",
    "\n",
    "\n",
    "def calculate_val_far(threshold, dist, actual_issame):\n",
    "    predict_issame = np.less(dist, threshold)\n",
    "    true_accept = np.sum(np.logical_and(predict_issame, actual_issame))\n",
    "    false_accept = np.sum(np.logical_and(predict_issame, np.logical_not(actual_issame)))\n",
    "    n_same = np.sum(actual_issame)\n",
    "    n_diff = np.sum(np.logical_not(actual_issame))\n",
    "    val = float(true_accept) / float(n_same)\n",
    "    far = float(false_accept) / float(n_diff) # how much percent predicted as positive, but actual negative(False positive rate)\n",
    "    return val, far\n",
    "\n",
    "\n",
    "\n",
    "def evaluate(embeddings, actual_issame, nrof_folds=10, distance_metric=0, subtract_mean=False):\n",
    "    thresholds = np.arange(0, 4, 0.01)\n",
    "    embeddings1 = embeddings[0::2]\n",
    "    embeddings2 = embeddings[1::2]\n",
    "    tpr, fpr, accuracy, fp, fn  = calculate_roc(thresholds, embeddings1, embeddings2,\n",
    "        np.asarray(actual_issame), nrof_folds=nrof_folds, distance_metric=distance_metric, subtract_mean=subtract_mean)\n",
    "    thresholds = np.arange(0, 4, 0.001)\n",
    "    val, val_std, far = calculate_val(thresholds, embeddings1, embeddings2,\n",
    "        np.asarray(actual_issame), 1e-3, nrof_folds=nrof_folds, distance_metric=distance_metric, subtract_mean=subtract_mean)\n",
    "    return tpr, fpr, accuracy, val, val_std, far, fp, fn\n",
    "\n",
    "\n",
    "def add_extension(path):\n",
    "    if os.path.exists(path+'.jpg'):\n",
    "        return path+'.jpg'\n",
    "    elif os.path.exists(path+'.png'):\n",
    "        return path+'.png'\n",
    "    else:\n",
    "        raise RuntimeError('No file \"%s\" with extension png or jpg.' % path)\n",
    "\n",
    "def get_paths(lfw_dir, pairs):\n",
    "    nrof_skipped_pairs = 0\n",
    "    path_list = []\n",
    "    issame_list = []\n",
    "    for pair in pairs:\n",
    "        if len(pair) == 3:\n",
    "            if pair[0] in names_without_image:\n",
    "                nrof_skipped_pairs += 1\n",
    "                continue\n",
    "            elif pair[0] in names_with_less_image:\n",
    "                pair1_int = min(len(name2img[pair[0]]), int(pair[1]))\n",
    "                pair2_int = min(len(name2img[pair[0]]), int(pair[2]))\n",
    "                path0 = add_extension(os.path.join(lfw_dir, pair[0], pair[0] + '_' + '%04d' % max(1, pair1_int-1) ))\n",
    "                path1 = add_extension(os.path.join(lfw_dir, pair[0], pair[0] + '_' + '%04d' % max(1, pair2_int-1) ))\n",
    "                issame = True\n",
    "            else:\n",
    "                path0 = add_extension(os.path.join(lfw_dir, pair[0], pair[0] + '_' + '%04d' % int(pair[1])))\n",
    "                path1 = add_extension(os.path.join(lfw_dir, pair[0], pair[0] + '_' + '%04d' % int(pair[2])))\n",
    "                issame = True\n",
    "        elif len(pair) == 4:\n",
    "            if pair[0] in names_without_image or pair[2] in names_without_image:\n",
    "                nrof_skipped_pairs += 1\n",
    "                continue\n",
    "            elif pair[0] in names_with_less_image:\n",
    "                pair1_int = min(len(name2img[pair[0]]), int(pair[1]))\n",
    "                path0 = add_extension(os.path.join(lfw_dir, pair[0], pair[0] + '_' + '%04d' % max(1, pair1_int-1) ))\n",
    "                path1 = add_extension(os.path.join(lfw_dir, pair[2], pair[2] + '_' + '%04d' % int(pair[3])))\n",
    "                issame = False\n",
    "            elif pair[2] in names_with_less_image:\n",
    "                pair3_int = min(len(name2img[pair[2]]), int(pair[3]))\n",
    "                path0 = add_extension(os.path.join(lfw_dir, pair[0], pair[0] + '_' + '%04d' % int(pair[1])))\n",
    "                path1 = add_extension(os.path.join(lfw_dir, pair[2], pair[2] + '_' + '%04d' % max(1, pair3_int-1) ))\n",
    "                issame = False\n",
    "            else:\n",
    "                path0 = add_extension(os.path.join(lfw_dir, pair[0], pair[0] + '_' + '%04d' % int(pair[1])))\n",
    "                path1 = add_extension(os.path.join(lfw_dir, pair[2], pair[2] + '_' + '%04d' % int(pair[3])))\n",
    "                issame = False\n",
    "        if os.path.exists(path0) and os.path.exists(path1):\n",
    "            path_list += [path0, path1]\n",
    "            issame_list.append(issame)\n",
    "        else:\n",
    "            nrof_skipped_pairs += 1\n",
    "    if nrof_skipped_pairs > 0:\n",
    "        print('Skipped %d image pairs' % nrof_skipped_pairs)\n",
    "    return path_list, issame_list\n",
    "\n",
    "\n",
    "def read_pairs(pairs_filename):\n",
    "    pairs = []\n",
    "    with open(pairs_filename, 'r') as f:\n",
    "        for line in f.readlines()[1:]:\n",
    "            pair = line.strip().split()\n",
    "            pairs.append(pair)\n",
    "    return np.array(pairs, dtype=object)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "b1815b9f",
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_embed_dict(bs, data_dir, device):\n",
    "    curr_idx = 0\n",
    "    path2idx = {}\n",
    "    embeddings = torch.tensor([])\n",
    "    faces = []\n",
    "    for name, imgs in name2img.items():\n",
    "        for img_idx, img in enumerate(imgs):\n",
    "            faces.append(img)\n",
    "            if name in names_with_less_image:\n",
    "                file_name = \"{}_{:04d}\".format(name, max(1, img_idx))\n",
    "            elif name in names_without_image:\n",
    "                continue\n",
    "            else:\n",
    "                file_name = \"{}_{:04d}\".format(name, img_idx+1)\n",
    "            path_dir = add_extension(os.path.join(data_dir, name, file_name))\n",
    "            path2idx[path_dir] = curr_idx\n",
    "            curr_idx += 1\n",
    "    num_batches = len(faces) // bs + int(len(faces) % bs > 0)\n",
    "    for batch in range(num_batches):\n",
    "        curr_batch = faces[batch*bs: (batch+1)*bs]\n",
    "        curr_batch = torch.stack(curr_batch, dim=0).to(device)\n",
    "        embeds = model(curr_batch).cpu().detach()\n",
    "        embeddings = torch.cat((embeddings, embeds), dim=0)\n",
    "    return embeddings, path2idx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "b36dbd31",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Skipped 17 image pairs\n"
     ]
    }
   ],
   "source": [
    "pairs = read_pairs(\"/archive/turganbay/data/lfw-py/lfw_funneled/pairs.txt\")\n",
    "path_list, issame_list = get_paths(\"/archive/turganbay/data/lfw-py/lfw_funneled\", pairs)\n",
    "embeddings, path2idx = create_embed_dict(4, \"/archive/turganbay/data/lfw-py/lfw_funneled/\", device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "e13cfd26",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True\n"
     ]
    }
   ],
   "source": [
    "embeddings_arr = [embeddings[path2idx[path]] for path in path_list]\n",
    "print( all([e.shape==torch.Size([512]) for e in embeddings_arr]) )\n",
    "embeddings_arr = np.array(torch.stack(embeddings_arr, dim=0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "c9388706",
   "metadata": {},
   "outputs": [],
   "source": [
    "tpr, fpr, accuracy, val, val_std, far, fp, fn = evaluate(embeddings_arr, issame_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "eadb2ab8",
   "metadata": {},
   "outputs": [],
   "source": [
    "accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "74cb6c54",
   "metadata": {},
   "source": [
    "## Evaluate using all the faces"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "b12cf131",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sentence_transformers import util\n",
    "\n",
    "def get_close_faces(query_face: torch.Tensor, embeddings: torch.Tensor, device):\n",
    "    inputs = query_face.unsqueeze(0).to(device)\n",
    "    query_embedding = model(inputs).cpu().detach()\n",
    "    \n",
    "    cos_scores = util.cos_sim(query_embedding, embeddings)[0]\n",
    "    top_results = torch.topk(cos_scores, k=1)\n",
    "    index = int(top_results[1].numpy())\n",
    "    top1_path = [k for k, v in path2idx.items() if v == index]\n",
    "    pred_name = top1_path[0].split(\"/\")[-2]\n",
    "    return pred_name\n",
    "\n",
    "\n",
    "def draw_pred_face(name: torch.Tensor, pred_lbl):\n",
    "    img_folder = os.path.join(\"./data/lfw-py/lfw_funneled\", name)\n",
    "    all_images_dir = os.listdir(img_folder)\n",
    "    images = []\n",
    "    for img in all_images_dir:\n",
    "        img_file = os.path.join(img_folder, img)\n",
    "        image = Image.open(img_file)\n",
    "        images.append(image)\n",
    "\n",
    "    number_of_subplots = len(images)\n",
    "    number_of_columns = 3\n",
    "    rows = number_of_subplots // number_of_columns \n",
    "    if number_of_subplots % number_of_columns != 0:\n",
    "        rows += 1\n",
    "\n",
    "    positions = range(1, number_of_subplots + 1)\n",
    "    plt.figure()\n",
    "    fig = plt.figure(1)\n",
    "    for subplot_num, (img, pred) in enumerate(zip(images, pred_lbl)):\n",
    "        ax = fig.add_subplot(rows, number_of_columns, positions[subplot_num])\n",
    "        ax.imshow(img)\n",
    "        ax.title.set_text(pred)\n",
    "        ax.set_xticks([])\n",
    "        ax.set_yticks([])\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "87ba048b",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Gold: Jessica_Capshaw\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "random_name = random.choice(names)\n",
    "name, num_images = random_name.split(\"\\t\")\n",
    "print(f\"Gold: {name}\")\n",
    "      \n",
    "preds = []\n",
    "for img in name2img[name]:\n",
    "    pred = get_close_faces(img, embeddings, device)\n",
    "    preds.append(pred)\n",
    "draw_pred_face(name, preds)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a6765ca0",
   "metadata": {},
   "source": [
    "### Some tips:\n",
    "- Quality of Training Data Matters Most: The quality of your training dataset is paramount. Ensure it is diverse, representative, and covers various age groups, ethnicities, poses, and lighting conditions. Low-quality or biased training data can severely impact your model's accuracy.\n",
    "\n",
    "- Data Augmentation for Robustness: Data augmentation techniques, such as random rotations, scaling, and brightness adjustments, can help improve the model's robustness. Experiment with different augmentations to make the model more adaptable to real-world scenarios.\n",
    "\n",
    "- Use Transfer Learning: Leveraging pre-trained models (e.g., from the ImageNet dataset) as a starting point can save time and improve performance. Fine-tune these models on your face recognition dataset to adapt them to your specific task.\n",
    "\n",
    "- Face Detection Preprocessing: Ensure effective face detection as a preprocessing step. Choose a robust face detection model (e.g., MTCNN or RetinaFace) to locate faces in images accurately.\n",
    "\n",
    "- Thresholding and Confidence Scores: Implement thresholding mechanisms and confidence scores to control the model's decisions. This can help reduce false positives or negatives in recognition tasks.\n",
    "\n",
    "- Liveness Detection: If security is a concern, consider incorporating liveness detection mechanisms to distinguish between real faces and spoof attempts, such as using a printed photo."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "env_llm",
   "language": "python",
   "name": "env_llm"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}