Spaces:
Running
Running
| import cv2 | |
| import streamlit as st | |
| import tempfile | |
| import torch | |
| from torchvision import transforms | |
| from mtcnn import MTCNN | |
| from skimage.feature import hog | |
| import joblib | |
| import numpy as np | |
| # Preprocessing for Siamese Model | |
| def preprocess_image_siamese(img): | |
| transform = transforms.Compose([ | |
| transforms.Resize((224, 224)), | |
| transforms.ToTensor() | |
| ]) | |
| img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) | |
| return transform(img) | |
| # Preprocessing for SVM model (converting to grayscale) | |
| def preprocess_image_svm(img): | |
| img = cv2.resize(img, (224, 224)) | |
| img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) | |
| return img | |
| # Extract HOG Features | |
| def extract_hog_features(img): | |
| hog_features = hog(img, orientations=9, pixels_per_cell=(16, 16), cells_per_block=(4, 4)) | |
| return hog_features | |
| # Detect faces using MTCNN | |
| def get_face(img): | |
| detector = MTCNN() | |
| faces = detector.detect_faces(img) | |
| if faces: | |
| x1, y1, w, h = faces[0]['box'] | |
| x1, y1 = abs(x1), abs(y1) | |
| x2, y2 = x1 + w, y1 + h | |
| return img[y1:y2, x1:x2] | |
| return None | |
| # Function to verify face (either HOG-SVM or Siamese model) | |
| def verify(img1, img2, model_type, anchor_img): | |
| with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_img1: | |
| temp_img1.write(img1.read()) | |
| temp_img1_path = temp_img1.name | |
| with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_img2: | |
| temp_img2.write(img2.read()) | |
| temp_img2_path = temp_img2.name | |
| img1p = cv2.imread(temp_img1_path) | |
| img2p = cv2.imread(temp_img2_path) | |
| face1 = get_face(img1p) | |
| face2 = get_face(img2p) | |
| if face1 is not None and face2 is not None: | |
| st.image([face1, face2], caption=["Image 1", "Image 2"], width=200) | |
| if model_type == "HOG-SVM": | |
| with open('./svm.pkl', 'rb') as f: | |
| svm = joblib.load(f) | |
| with open('./pca.pkl', 'rb') as f: | |
| pca = joblib.load(f) | |
| face1 = preprocess_image_svm(face1) | |
| face2 = preprocess_image_svm(face2) | |
| hog1 = extract_hog_features(face1) | |
| hog2 = extract_hog_features(face2) | |
| hog1_pca = pca.transform([hog1]) | |
| hog2_pca = pca.transform([hog2]) | |
| pred1 = svm.predict(hog1_pca) | |
| pred2 = svm.predict(hog2_pca) | |
| if pred1 == 1 and pred2 == 1: | |
| st.write("Matched") | |
| else: | |
| st.write("Not Matched") | |
| else: | |
| st.write("Face not detected in one or both images") | |
| # Main function to handle Streamlit interaction | |
| def main(): | |
| st.title("Real-time Face Verification App") | |
| model_type = st.selectbox("Select Model", ["Siamese", "HOG-SVM"]) | |
| anchor_img = st.file_uploader("Select Anchor Image", type=["jpg", "png"]) | |
| if model_type == "Siamese": | |
| # Implement Siamese model choice logic here if needed | |
| st.write("Using Siamese Network") | |
| elif model_type == "HOG-SVM": | |
| # Implement HOG-SVM logic here | |
| st.write("Using HOG-SVM") | |
| # Camera Input for Face Detection | |
| run_detection = st.checkbox("Start Camera") | |
| if run_detection: | |
| cap = cv2.VideoCapture(0) # Start camera | |
| while True: | |
| ret, frame = cap.read() | |
| if not ret: | |
| st.write("Failed to grab frame.") | |
| break | |
| # Detect face in the current frame | |
| face = get_face(frame) | |
| if face is not None: | |
| # Draw bounding box around detected face | |
| x1, y1, x2, y2 = face[0], face[1], face[2], face[3] # Update face coordinates | |
| cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2) | |
| # Show bounding box | |
| st.image(frame, channels="BGR", use_column_width=True) | |
| # Stop camera when ESC is pressed | |
| key = cv2.waitKey(1) & 0xFF | |
| if key == 27: # ESC key | |
| break | |
| cap.release() | |
| cv2.destroyAllWindows() | |
| if __name__ == "__main__": | |
| main() | |