Augmentation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from PIL import Image\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from skimage.util import random_noise"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "DIRECTROY = 'data'\n",
    "MODEL_PATH = 'models'\n",
    "IMG_SIZE = 224"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def resize_img(a):\n",
    "    w, h = a.size\n",
    "    if w < h:\n",
    "        scale = h/IMG_SIZE\n",
    "        w = int(w/scale)\n",
    "        h = IMG_SIZE\n",
    "        a = a.resize((w, h))\n",
    "        lb= np.array([a.load()[0,x] for x in range(h)])\n",
    "        rb = np.array([a.load()[w-1,x] for x in range(h)])\n",
    "        lb = lb.mean(axis=0).astype('uint8')\n",
    "        rb = rb.mean(axis=0).astype('uint8')\n",
    "        pic = Image.new('RGB', (h, h), color = (255, 255, 255))\n",
    "        imgl = Image.new('RGB', (h//2, h), color = tuple(lb))\n",
    "        imgr = Image.new('RGB', (h//2, h), color = tuple(rb))\n",
    "        \n",
    "        pic.paste(imgl, (0, 0))\n",
    "        pic.paste(imgr, (h//2, 0))\n",
    "        \n",
    "        pic.paste(a, (h//2-w//2, 0))\n",
    "\n",
    "    elif w>h:\n",
    "        scale = w/IMG_SIZE\n",
    "        h = int(h/scale)\n",
    "        w = IMG_SIZE\n",
    "        a = a.resize((w, h))\n",
    "        \n",
    "        lb= np.array([a.load()[x,0] for x in range(w)])\n",
    "        rb = np.array([a.load()[x,h-1] for x in range(w)])\n",
    "        lb = lb.mean(axis=0).astype('uint8')\n",
    "        rb = rb.mean(axis=0).astype('uint8')\n",
    "        \n",
    "        pic = Image.new('RGB', (w, w), color = (255, 255, 255))\n",
    "        imgl = Image.new('RGB', (w, w//2), color = tuple(lb))\n",
    "        imgr = Image.new('RGB', (w, w//2), color = tuple(rb))\n",
    "        \n",
    "        pic.paste(imgl, (0, 0))\n",
    "        pic.paste(imgr, (0, w//2))\n",
    "        \n",
    "        pic.paste(a, (0, w//2-h//2))\n",
    "\n",
    "    else:\n",
    "        a = a.resize((IMG_SIZE, IMG_SIZE))\n",
    "        pic = a\n",
    "    return pic"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Change image in following methods\n",
    "\n",
    "- Vertical flip\n",
    "- Horizontal flip\n",
    "- Rotate 30 degree\n",
    "- Zoom\n",
    "- Random noise\n",
    "- Random blocks cover the images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train = pd.read_csv(f'{DIRECTROY}/reduced_train.csv') \n",
    "df_test = pd.read_csv(f'{DIRECTROY}/reduced_test.csv') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def flip_image_vertical(image):\n",
    "    return Image.fromarray(np.fliplr(image))\n",
    "\n",
    "def flip_image_horizontal(image):\n",
    "    return Image.fromarray(np.flipud(image))\n",
    "\n",
    "def rotate_image(image, angle=30):\n",
    "    return image.rotate(angle)\n",
    "\n",
    "\n",
    "def randomCrop(im, margin=1.25):\n",
    "    '''\n",
    "    croping the image in the center from a random margin from the borders\n",
    "    '''\n",
    "    mg = int(np.random.rand()*30)\n",
    "    margin -=1\n",
    "    margin/=2\n",
    "    start = [mg + im.size[0]*margin, mg + im.size[0]*margin]\n",
    "    end = [min(mg+ im.size[0] * (1-margin), im.size[0]),  min(mg+im.size[1]*(1-margin), im.size[1])]\n",
    "    cropped_image = im.crop((start[0], start[1], end[0], end[1]))\n",
    "    return cropped_image\n",
    "\n",
    "def add_noise(image):\n",
    "    img =  random_noise(np.array(image))*255\n",
    "    img = Image.fromarray(img.astype('uint8'))\n",
    "    return img\n",
    "\n",
    "def add_random_box(img):\n",
    "    box_width = np.random.randint(20,50)\n",
    "    color = list(np.random.choice(range(256), size=3))\n",
    "    img2 = img.copy()\n",
    "    img2.paste(Image.new('RGB', (box_width, box_width), color = tuple(color)), (np.random.randint(0,IMG_SIZE-box_width), np.random.randint(0,IMG_SIZE-box_width)))\n",
    "    return img2\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def resize_images(df,tasks='train'):\n",
    "    for x in df['name']:\n",
    "        img = Image.open(f'{DIRECTROY}/{tasks}/{x}')\n",
    "        img = resize_img(img)\n",
    "        img.save(f'{DIRECTROY}/aug_{tasks}/{x}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "resize_images(df_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "img = Image.open(f'{DIRECTROY}/train/1.jpg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "img = resize_img(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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",
      "image/png": "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",
      "text/plain": [
       "<PIL.Image.Image image mode=RGB size=224x224>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "flip_image_horizontal(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tqdm import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def augment_image(df_train, type='horizontal'):\n",
    "    df_aug = df_train.sample(frac=0.245)\n",
    "    images = df_aug['name'].values\n",
    "    for image in tqdm(images, desc=f'Augmenting {type} images'):\n",
    "        im = Image.open(f'{DIRECTROY}/train/{image}')\n",
    "        if type == 'horizontal':\n",
    "            im = flip_image_horizontal(im)\n",
    "        elif type == 'vertical':\n",
    "            im = flip_image_vertical(im)\n",
    "        elif type == 'rotate':\n",
    "            im = resize_img(im)\n",
    "            im = rotate_image(im)\n",
    "        elif type == 'crop':\n",
    "            im = resize_img(im)\n",
    "            im = randomCrop(im)\n",
    "        elif type == 'noise':\n",
    "            im = add_noise(im)\n",
    "        elif type == 'box':\n",
    "            im = resize_img(im)\n",
    "            im = add_random_box(im)\n",
    "        im = resize_img(im)\n",
    "        im.save(f'{DIRECTROY}/aug_train/{type}_{image}')\n",
    "    \n",
    "    df_aug['name'] = df_aug['name'].apply(lambda x: f'{type}_{x}')\n",
    "    return df_aug"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Augmenting horizontal images:   0%|          | 0/7165 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Augmenting horizontal images: 100%|██████████| 7165/7165 [02:45<00:00, 43.38it/s]\n",
      "Augmenting vertical images: 100%|██████████| 7165/7165 [02:37<00:00, 45.37it/s]\n",
      "Augmenting rotate images: 100%|██████████| 7165/7165 [01:29<00:00, 79.70it/s] \n",
      "Augmenting crop images: 100%|██████████| 7165/7165 [01:22<00:00, 87.07it/s] \n",
      "Augmenting noise images: 100%|██████████| 7165/7165 [10:31<00:00, 11.34it/s]\n",
      "Augmenting box images: 100%|██████████| 7165/7165 [01:23<00:00, 85.49it/s] \n"
     ]
    }
   ],
   "source": [
    "df_h = augment_image(df_train, type='horizontal')\n",
    "df_v = augment_image(df_train, type='vertical')\n",
    "df_r = augment_image(df_train, type='rotate')\n",
    "df_c = augment_image(df_train, type='crop')\n",
    "df_n = augment_image(df_train, type='noise')\n",
    "df_b = augment_image(df_train, type='box')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(len(df_train)):\n",
    "    img = Image.open(f'{DIRECTROY}/train/{df_train.iloc[i].name}')\n",
    "    img = resize_img(img)\n",
    "    img.save(f'{DIRECTROY}/resize_train/{df_train.iloc[i].name}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_aug = pd.concat([df_train, df_h, df_v, df_r, df_c, df_n,df_b])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "72233"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(df_aug)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "29243"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(df_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_aug.sample(frac=1).to_csv(f'{DIRECTROY}/augmented_train.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_aug = pd.read_csv(f'{DIRECTROY}/augmented_train.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create dataloader \n",
    "Split dataset to multiple pack due to lack of ram"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import CLIPTokenizerFast\n",
    "import torch\n",
    "from torchvision.transforms import Compose, ToTensor, Lambda, Resize, Normalize\n",
    "from torch.utils.data import Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "image_transforms = Compose([\n",
    "    Resize((IMG_SIZE, IMG_SIZE)),\n",
    "    ToTensor(), \n",
    "    Normalize(mean=[0.48145466, 0.4578275, 0.40821073], std=[0.26862954, 0.26130258, 0.27577711])\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "class CustomDataset(Dataset):\n",
    "    def __init__(self, df, transforms, directory):\n",
    "        self.tokenizer =  CLIPTokenizerFast.from_pretrained(\"openai/clip-vit-base-patch16\")\n",
    "        self.df = df\n",
    "        self.transforms = transforms\n",
    "        self.directory = directory\n",
    "        self.labels = torch.Tensor(df['newid'].values).long()\n",
    "        self.imgs = torch.cat([ self.transforms(Image.open(f'{DIRECTROY}/{self.directory}/{x}').convert('RGB')).half().reshape(1,3,IMG_SIZE,IMG_SIZE) for x in tqdm(df['name'].values)])\n",
    "        self.tokenized = self.tokenizer([f'a photo of {x}' for x in df['label'].values], padding=True, truncation=True, return_tensors=\"pt\")\n",
    "        self.input_ids = self.tokenized['input_ids']\n",
    "        self.attention_mask = self.tokenized['attention_mask']\n",
    "        \n",
    "    def __len__(self):\n",
    "        return len(self.df)\n",
    "\n",
    "    def __getitem__(self, idx):\n",
    "        img = self.imgs[idx]\n",
    "        label = self.labels[idx]\n",
    "        input_ids = self.input_ids[idx]\n",
    "        attention_mask = self.attention_mask[idx]\n",
    "        return img, label, input_ids, attention_mask"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "import math"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 36384/36384 [00:26<00:00, 1357.83it/s]\n",
      "100%|██████████| 35849/35849 [01:44<00:00, 341.93it/s]\n"
     ]
    }
   ],
   "source": [
    "for i in range(math.ceil(len(df_aug)/36384)):\n",
    "    train_dataset = CustomDataset(df_aug[i*36384:(i+1)*36384], image_transforms, 'aug_train')\n",
    "    torch.save(train_dataset, f'{DIRECTROY}/train_dataset/train_dataset_reduced_aug_{i}.pth')\n",
    "    del train_dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_dataset = CustomDataset(df_test, image_transforms, 'test')\n",
    "torch.save(test_dataset, f'{DIRECTROY}/test_public_dataset/test_dataset_reduced_2.pth')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",

   "version": "3.10.11"

  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}