Update app.py

app.py

@@ -1,58 +1,46 @@
 import gradio as gr
-import cv2
 import numpy as np
+import cv2
 import os
 
-# Load your trained model and database here
-# Example: load mean face, eigenfaces, and face projections
+# Load model data
 mean = np.load("mean.npy")
 eigenfaces = np.load("eigenfaces.npy")
 projected_faces = np.load("projected_faces.npy")
-labels = np.load("labels.npy")  # Assuming you saved labels for your dataset
-
-def preprocess_image(image):
-    # Convert to grayscale
-    gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
-    resized = cv2.resize(gray, (100, 100)).flatten()
-    normalized = resized.astype('float32')
-    return normalized
-
-def project_face(face_vector, mean, eigenfaces):
-    centered_vector = face_vector - mean
-    return np.dot(eigenfaces, centered_vector)
-
-def recognize_face(uploaded_image):
+labels = np.load("labels.npy")
+
+# Preprocessing
+def preprocess(img):
+    img = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
+    img = cv2.resize(img, (100, 100))  # or your dataset size
+    img_flat = img.flatten()
+    return img_flat
+
+# Prediction function
+def predict(img):
+    if img is None:
+        return "No image provided"
+    
     try:
-        if uploaded_image is None:
-            return "No image provided."
-
-        face_vector = preprocess_image(uploaded_image)
-        projected = project_face(face_vector, mean, eigenfaces)
-
-        # Compare with database
-        min_dist = float('inf')
-        matched_label = "Unknown"
-        for idx, known_proj in enumerate(projected_faces):
-            dist = np.linalg.norm(projected - known_proj)
-            if dist < min_dist:
-                min_dist = dist
-                matched_label = labels[idx]
-
-        threshold = 3000  # Adjust based on your model
-        if min_dist < threshold:
-            return f"Matched: {matched_label} (Distance: {min_dist:.2f})"
-        else:
-            return "No match found"
+        img_vector = preprocess(img)
+        img_centered = img_vector - mean
+        img_projected = np.dot(eigenfaces.T, img_centered)
+
+        distances = np.linalg.norm(projected_faces - img_projected, axis=1)
+        min_index = np.argmin(distances)
+        predicted_label = labels[min_index]
+        
+        return f"Match found: Person {predicted_label} (distance={distances[min_index]:.2f})"
     except Exception as e:
-        return f"Error: {str(e)}"
+        return f"Error processing image: {str(e)}"
 
+# Gradio UI
 iface = gr.Interface(
-    fn=recognize_face,
-    inputs=gr.Image(type="numpy", label="Take or upload a photo"),
-    outputs=gr.Textbox(label="Prediction"),
-    title="Eigenface Face Recognition",
-    description="Upload or take a photo. The model will identify the person if in the database.",
-    allow_flagging="never",
+    fn=predict,
+    inputs=gr.Image(type="numpy", label="Upload Face Image"),
+    outputs=gr.Textbox(label="Prediction Result"),
+    title="Eigenfaces Face Recognition",
+    description="Upload a face image to identify the most similar person from the dataset."
 )
 
 iface.launch()