Update app.py

app.py

@@ -1,11 +1,10 @@
-# SecureFace ID – FINAL VERSION THAT WORKS EVERYWHERE (Nov 2025)
-
+# SecureFace ID – DEBUG & FIXED VERSION (Threshold 0.6 + Logging)
 import os
 import cv2
 import numpy as np
 import gradio as gr
 from ultralytics import YOLO
-from huggingface_hub import hf_hub_download          # ← fixed import
+from huggingface_hub import hf_hub_download
 import insightface
 from insightface.app import FaceAnalysis
 import faiss
@@ -16,18 +15,15 @@ KNOWN_EMBS_PATH = "known_embeddings.npy"
 KNOWN_NAMES_PATH = "known_names.npy"
 
 # ==================== MODELS ====================
-# YOLOv8 face detector – auto-downloaded first run
 model_path = hf_hub_download(
     repo_id="arnabdhar/YOLOv8-Face-Detection",
     filename="model.pt"
 )
 detector = YOLO(model_path)
 
-# InsightFace buffalo_l (best model 2025)
 recognizer = FaceAnalysis(name='buffalo_l', providers=['CPUExecutionProvider'])
 recognizer.prepare(ctx_id=0, det_size=(640, 640))
 
-# Tracker + FAISS index
 tracker = DeepSort(max_age=30, n_init=3, max_cosine_distance=0.4, embedder_gpu=False)
 index = faiss.IndexHNSWFlat(512, 32)
 index.hnsw.efSearch = 16
@@ -35,18 +31,18 @@ known_names = []
 unknown_counter = 0
 track_to_label = {}
 
-# Load existing database at startup
+# Load database at startup
 if os.path.exists(KNOWN_EMBS_PATH) and os.path.getsize(KNOWN_EMBS_PATH) > 0:
     embs = np.load(KNOWN_EMBS_PATH)
     known_names = np.load(KNOWN_NAMES_PATH, allow_pickle=True).tolist()
     index.add(embs.astype('float32'))
     print(f"Loaded {len(known_names)} people")
 
-# ==================== PROCESS FRAME ====================
+# ==================== PROCESS FRAME (WITH DEBUG LOGGING) ====================
 def process_frame(frame, blur_type="gaussian", intensity=40, expand=1.3, show_labels=True):
     global unknown_counter, track_to_label
 
-    img = frame.copy()
+    img = frame.copy()  # Gradio gives RGB
     h, w = img.shape[:2]
     results = detector(img, conf=0.4)[0]
 
@@ -70,12 +66,16 @@ def process_frame(frame, blur_type="gaussian", intensity=40, expand=1.3, show_la
         tid = track.track_id
 
         if tid not in track_to_label or track.time_since_update % 15 == 0:
-            faces = recognizer.get(crop, max_num=1)
+            # Convert to BGR for InsightFace (if needed)
+            crop_bgr = cv2.cvtColor(crop, cv2.COLOR_RGB2BGR)
+            faces = recognizer.get(crop_bgr, max_num=1)
             name = "Unknown"
             if faces and index.ntotal > 0:
                 emb = faces[0].normed_embedding.reshape(1, -1).astype('float32')
                 D, I = index.search(emb, 1)
-                if D[0][0] < 0.45:
+                distance = D[0][0]
+                print(f"DEBUG: Distance = {distance:.3f} for potential match to index {I[0][0]} ({known_names[I[0][0]] if I[0][0] < len(known_names) else 'invalid'})")
+                if distance < 0.6:  # ← FIXED: More lenient threshold
                     name = known_names[I[0][0]]
 
             if name == "Unknown":
@@ -88,7 +88,7 @@ def process_frame(frame, blur_type="gaussian", intensity=40, expand=1.3, show_la
 
         label = track_to_label[tid]
 
-        # Blur face
+        # Blur
         face = img[y1:y2, x1:x2]
         if blur_type == "gaussian":
             k = max(15, int(min(x2-x1, y2-y1) * intensity / 100) | 1)
@@ -106,20 +106,22 @@ def process_frame(frame, blur_type="gaussian", intensity=40, expand=1.3, show_la
 
     return img
 
-# ==================== ENROLL ====================
+# ==================== ENROLL (SAME PREPROCESSING) ====================
 def enroll_person(name, face_image):
     global index, known_names
 
     if face_image is None or name.strip() == "":
         return "Add name + photo"
 
-    faces = recognizer.get(face_image, max_num=1)
+    # Convert to BGR for consistency
+    face_bgr = cv2.cvtColor(face_image, cv2.COLOR_RGB2BGR)
+    faces = recognizer.get(face_bgr, max_num=1)
     if not faces:
-        return "No face detected – try again"
+        return "No face detected"
 
     new_emb = faces[0].normed_embedding.reshape(1, 512)
 
-    # Load or create database
+    # Load or create
     if os.path.exists(KNOWN_EMBS_PATH) and os.path.getsize(KNOWN_EMBS_PATH) > 0:
         embs = np.load(KNOWN_EMBS_PATH)
         names = np.load(KNOWN_NAMES_PATH, allow_pickle=True).tolist()
@@ -130,13 +132,13 @@ def enroll_person(name, face_image):
     embs = np.vstack([embs, new_emb])
     names.append(name)
 
-    # Save & update index
     np.save(KNOWN_EMBS_PATH, embs)
     np.save(KNOWN_NAMES_PATH, np.array(names))
     index.reset()
     index.add(embs.astype('float32'))
     known_names = names
 
+    print(f"ENROLL DEBUG: Added {name} with embedding shape {new_emb.shape}")
     return f"**{name}** enrolled successfully!"
 
 # ==================== UI ====================