Merge branch 'develop' into ljy67122

.gitignore

@@ -3,3 +3,4 @@ wider face.v3i.yolov5pytorch
 dist
 build
 myenv
+svm/**

pytorch/mosaic_jiyeon.py

@@ -7,7 +7,7 @@
 def mosaic(video_path, image_paths):
     # YOLOv5 모델 로드
     # model = torch.hub.load('./pytorch/yolov5', 'custom', path='./pytorch/best.pt', source='local')
-    model = torch.hub.load('ultralytics/yolov5', 'custom', path='./pytorch/best.pt')
+    model = torch.hub.load('ultralytics/yolov5', 'custom', path='./pytorch/best.pt',source='local')
     output_video_path = os.path.join('tmp', video_path)
     # 특정 사람의 얼굴 이미지 로드
     # person_image = face_recognition.load_image_file("goognyoo.png")

svm/cnn.py

@@ -0,0 +1,82 @@
+import os
+import cv2
+import numpy as np
+from tensorflow.keras import Sequential
+from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout, Input
+from tensorflow.keras.preprocessing.image import ImageDataGenerator
+from sklearn.model_selection import train_test_split
+from tensorflow.keras.utils import to_categorical
+
+# 이미지 로딩 및 전처리 함수
+def load_images_from_directory(directory, size=(64, 64)):
+    images = []
+    for filename in os.listdir(directory):
+        img_path = os.path.join(directory, filename)
+        if filename.endswith(".jpg") or filename.endswith(".png"):
+            img = cv2.imread(img_path)
+            if img is None:
+                print(f"Warning: Image not loaded properly from {img_path}")
+                continue
+            img = cv2.resize(img, size)
+            img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+            images.append(img)
+
+    # Convert the list of images to a 4D numpy array
+    images = np.array(images)  # This should create an array of shape (num_images, 64, 64, 3)
+    return images
+
+# 데이터셋 로드
+# celeba_images = load_images_from_directory('img_align_celeba')
+images = load_images_from_directory('output')
+
+print("Data Loaded Successfully")
+# print(f"Number of CelebA images: {len(celeba_images)}")
+print(f"Number of Output images: {len(images)}")
+
+# 데이터 결합 및 분할
+all_images = np.concatenate(( images), axis=0)
+X_train, X_test = train_test_split(images, test_size=0.2, random_state=42)
+
+data_generator = ImageDataGenerator(rotation_range=20, width_shift_range=0.2, height_shift_range=0.2, horizontal_flip=True)
+train_generator = data_generator.flow(X_train, batch_size=32)
+
+
+
+print("Data split into training and test sets.")
+print(f"Training set size: {len(X_train)}")
+print(f"Test set size: {len(X_test)}")
+
+# 모델 구축 함수
+def build_model(input_shape):
+    model = Sequential([
+        Input(shape=input_shape),
+        Conv2D(32, (3, 3), activation='relu'),
+        MaxPooling2D((2, 2)),
+        Conv2D(64, (3, 3), activation='relu'),
+        MaxPooling2D((2, 2)),
+        Conv2D(128, (3, 3), activation='relu'),
+        MaxPooling2D((2, 2)),
+        Flatten(),
+        Dense(128, activation='relu'),
+        Dropout(0.5),
+        Dense(2, activation='softmax')  # Assuming binary classification for simplicity
+    ])
+    model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])
+    return model
+
+model = build_model((64, 64, 3))
+
+# 데이터 증강
+data_generator = ImageDataGenerator(rotation_range=20, width_shift_range=0.2, height_shift_range=0.2, horizontal_flip=True)
+train_generator = data_generator.flow(X_train, batch_size=32)
+
+# 모델 학습 (임시로 labels 만들기 - 실제 사용 시 적절한 labels 준비 필요)
+dummy_labels = np.random.randint(2, size=len(X_train))
+dummy_labels = to_categorical(dummy_labels, num_classes=2)
+# Assuming you have a 'model' defined
+model.fit(train_generator, epochs=10, validation_data=(X_test, np.random.randint(0, 2, size=(len(X_test), 1))))
+print("Model training complete.")
+
+# 모델 저장
+model.save('face_recognition_model.h5')
+print("Model saved successfully.")

svm/cnn2.py

@@ -0,0 +1,68 @@
+import cv2
+import numpy as np
+from skimage.feature import hog
+from sklearn.model_selection import train_test_split, GridSearchCV
+from sklearn import svm
+from sklearn.metrics import classification_report, confusion_matrix
+import joblib
+import os
+
+# 특징 추출 함수
+def extract_hog_features(image_path):
+    image = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
+    if image is None:
+        print(f"Failed to load image: {image_path}")
+        return None
+    resized_img = cv2.resize(image, (64, 64))
+    fd = hog(resized_img, orientations=8, pixels_per_cell=(16, 16),
+             cells_per_block=(1, 1), visualize=False, feature_vector=True)
+    return fd
+
+def load_images_from_directory(directory, sample_size=None):
+    features = []
+    labels = []
+    image_files = [os.path.join(directory, file) for file in os.listdir(directory) if file.endswith('.jpg')]
+    if sample_size:
+        image_files = np.random.choice(image_files, size=sample_size, replace=False)
+
+    total_images = len(image_files)
+    for index, file in enumerate(image_files):
+        feat = extract_hog_features(file)
+        if feat is not None:
+            features.append(feat)
+            labels.append(1)  # 모든 이미지에 대해 '얼굴' 레이블(1) 부여
+        print(f"Processed {index + 1}/{total_images} images.")
+
+    return features, labels
+
+# 데이터셋 경로 설정
+celeba_dir = './img_align_celeba'  # CelebA 이미지 디렉토리
+wider_face_dir = './output'  # WIDER FACE 이미지 디렉토리
+
+# Load data
+celeba_features, celeba_labels = load_images_from_directory(celeba_dir, sample_size=500)
+wider_features, wider_labels = load_images_from_directory(wider_face_dir, sample_size=500)
+
+# 데이터 합치기
+features = np.vstack((celeba_features, wider_features))
+labels = np.hstack((celeba_labels + wider_labels))
+print('데이터 합치기 끝')
+
+# 데이터 분할
+X_train, X_test, y_train, y_test = train_test_split(features, labels, test_size=0.25, random_state=42)
+print('데이터 분할 끝')
+
+# 그리드 서치를 이용한 SVM 튜닝 및 훈련
+parameters = {'C': [0.1, 1, 10, 100], 'gamma': [1, 0.1, 0.01, 0.001], 'kernel': ['rbf', 'linear']}
+grid_search = GridSearchCV(svm.SVC(), parameters, refit=True, verbose=2, cv=3)
+grid_search.fit(X_train, y_train)
+print('튜닝 및 훈련 끝')
+
+# 모델 평가
+y_pred = grid_search.predict(X_test)
+print("Best parameters found:", grid_search.best_params_)
+print("Classification Report:\n", classification_report(y_test, y_pred))
+print("Confusion Matrix:\n", confusion_matrix(y_test, y_pred))
+
+# 모델 저장
+joblib.dump(grid_search.best_estimator_, 'svm_face_recognition2.pkl')