Spaces:
Sleeping
Sleeping
| import os | |
| import cv2 | |
| import torch | |
| import numpy as np | |
| import streamlit as st | |
| import requests | |
| from PIL import Image | |
| from glob import glob | |
| from insightface.app import FaceAnalysis | |
| import torch.nn.functional as F | |
| # Set the device | |
| device = torch.device("cuda" if torch.cuda.is_available() else "cpu") | |
| # Global Variables | |
| IMAGE_SHAPE = 640 | |
| data_path = 'employees' | |
| webcam_path = 'captured_image.jpg' | |
| # Set Streamlit title | |
| st.title("📸 AI-Driven Multi-Face Attendance System") | |
| # Load employee image paths | |
| image_paths = glob(os.path.join(data_path, '*.jpg')) | |
| employee_names = [os.path.basename(p).split('.')[0] for p in image_paths] | |
| # Initialize Face Analysis | |
| app = FaceAnalysis(name="buffalo_l") # ArcFace model | |
| app.prepare(ctx_id=0 if torch.cuda.is_available() else -1, det_size=(IMAGE_SHAPE, IMAGE_SHAPE)) | |
| # Load all employee embeddings | |
| employee_embeddings = [] | |
| for path in image_paths: | |
| img = cv2.imread(path) | |
| faces = app.get(img, max_num=1) | |
| if faces: | |
| embedding = torch.tensor(faces[0].embedding, dtype=torch.float32) | |
| employee_embeddings.append(embedding) | |
| else: | |
| employee_embeddings.append(None) | |
| # Function to compute cosine similarity and find best match | |
| def find_best_match(face_emb): | |
| similarities = [] | |
| for emp_emb in employee_embeddings: | |
| if emp_emb is None: | |
| similarities.append(torch.tensor(-1.0)) | |
| continue | |
| score = F.cosine_similarity(emp_emb, face_emb, dim=0) | |
| similarities.append(score) | |
| scores = torch.stack(similarities) | |
| best_idx = torch.argmax(scores) | |
| return scores[best_idx].item(), best_idx | |
| # Streamlit tabs | |
| about_tab, app_tab = st.tabs(["About the app", "Face Recognition"]) | |
| with about_tab: | |
| st.markdown(""" | |
| # 🎯 Multi-Face AI Attendance System | |
| - Detects multiple faces in one frame | |
| - Matches using ArcFace embeddings | |
| - Marks attendance for each recognized student | |
| """) | |
| st.markdown("### 📚 Students Trained in the System:") | |
| if employee_names: | |
| for name in employee_names: | |
| st.markdown(f"- {name}") | |
| else: | |
| st.warning("No student images found in the 'employees/' folder.") | |
| with app_tab: | |
| enable = st.checkbox("Enable camera") | |
| picture = st.camera_input("Take a picture", disabled=not enable) | |
| if picture is not None: | |
| with st.spinner("Analyzing faces..."): | |
| image_pil = Image.open(picture) | |
| image_np = np.array(image_pil) | |
| image_bgr = cv2.cvtColor(image_np, cv2.COLOR_RGB2BGR) | |
| detected_faces = app.get(image_bgr) | |
| if not detected_faces: | |
| st.warning("No face detected in the captured image.") | |
| else: | |
| marked_names = [] | |
| for face in detected_faces: | |
| emb = torch.tensor(face.embedding, dtype=torch.float32) | |
| score, idx = find_best_match(emb) | |
| if score >= 0.6: | |
| name = employee_names[idx] | |
| # Draw box and name | |
| x1, y1, x2, y2 = [int(i) for i in face.bbox] | |
| cv2.rectangle(image_bgr, (x1, y1), (x2, y2), (0, 255, 0), 2) | |
| cv2.rectangle(image_bgr, (x1, y2 - 35), (x2, y2), (0, 255, 0), cv2.FILLED) | |
| cv2.putText(image_bgr, name, (x1 + 6, y2 - 10), cv2.FONT_HERSHEY_COMPLEX, 0.8, (255, 255, 255), 2) | |
| # Mark attendance only once | |
| if name not in marked_names: | |
| marked_names.append(name) | |
| # Send attendance via POST | |
| url = "https://attendance-system-25.glitch.me/adds" | |
| data = {'rno': 15, 'sname': name, 'sclass': 7} | |
| try: | |
| response = requests.post(url, data=data) | |
| if response.status_code == 200: | |
| st.success(f"✅ Attendance marked for {name}") | |
| else: | |
| st.warning(f"⚠️ Attendance failed for {name}") | |
| except Exception as e: | |
| st.error(f"Request failed for {name}: {e}") | |
| else: | |
| x1, y1, x2, y2 = [int(i) for i in face.bbox] | |
| cv2.rectangle(image_bgr, (x1, y1), (x2, y2), (0, 0, 255), 2) | |
| cv2.putText(image_bgr, "Unknown", (x1 + 6, y2 - 10), cv2.FONT_HERSHEY_COMPLEX, 0.8, (255, 255, 255), 2) | |
| # Show final image | |
| final_image = cv2.cvtColor(image_bgr, cv2.COLOR_BGR2RGB) | |
| st.image(final_image, caption="Detected Faces with Names", use_column_width=True) | |