Update app.py

app.py

@@ -1,76 +1,61 @@
 import gradio as gr
 import cv2
 import numpy as np
+import tempfile
+from PIL import Image
 import mediapipe as mp
-import uuid
-import os
 from sklearn.cluster import KMeans
 
-# Create reusable face mesh
+# MediaPipe setup
 mp_face_mesh = mp.solutions.face_mesh
 face_mesh = mp_face_mesh.FaceMesh(static_image_mode=True)
+mp_drawing = mp.solutions.drawing_utils
 
-# Face Shape Detection Logic (Simplified based on face landmarks)
-def detect_face_shape(landmarks, image_shape):
-    # Grab required landmark points
-    left_cheek = landmarks[234]
-    right_cheek = landmarks[454]
-    chin = landmarks[152]
-    forehead = landmarks[10]
-
-    # Calculate distances
-    width = np.linalg.norm(np.array([left_cheek.x, left_cheek.y]) - np.array([right_cheek.x, right_cheek.y]))
-    height = np.linalg.norm(np.array([chin.x, chin.y]) - np.array([forehead.x, forehead.y]))
-
-    ratio = width / height
-    if ratio > 1.3:
-        return "Round"
-    elif ratio > 1.1:
-        return "Oval"
-    else:
-        return "Long"
-
-# Skin tone using KMeans
-def get_skin_tone(image):
-    h, w, _ = image.shape
-    face_crop = image[h//4:3*h//4, w//3:2*w//3]  # middle region
-
-    pixels = face_crop.reshape(-1, 3)
-    kmeans = KMeans(n_clusters=3, random_state=0).fit(pixels)
-    dominant = kmeans.cluster_centers_.astype(int)[0]
-    return tuple(dominant)
-
+# Step 1 & 2 combined: Capture, analyze, and return image with annotations
 def analyze_face(image):
     if image is None:
-        return None, "No image", "No skin tone"
-    
-    image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
-    results = face_mesh.process(image_rgb)
-
-    if not results.multi_face_landmarks:
-        return image, "No face detected", "N/A"
-
-    landmarks = results.multi_face_landmarks[0].landmark
-    shape = detect_face_shape(landmarks, image.shape)
-
-    tone = get_skin_tone(image)
-    tone_str = f"RGB: {tone}"
-
-    return image, shape, tone_str
-
-# Gradio interface
+        return "No image provided"
+
+    # Save image temporarily
+    temp_file = tempfile.NamedTemporaryFile(suffix=".png", delete=False)
+    image_pil = Image.fromarray(image)
+    image_pil.save(temp_file.name)
+
+    # Convert for MediaPipe
+    img_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+    result = face_mesh.process(img_rgb)
+
+    if not result.multi_face_landmarks:
+        return "No face detected"
+
+    # Draw landmarks
+    for face_landmarks in result.multi_face_landmarks:
+        mp_drawing.draw_landmarks(
+            image=image,
+            landmark_list=face_landmarks,
+            connections=mp_face_mesh.FACEMESH_TESSELATION,
+            landmark_drawing_spec=None,
+            connection_drawing_spec=mp_drawing.DrawingSpec(color=(0,255,0), thickness=1, circle_radius=1),
+        )
+
+    # Skin tone detection with KMeans
+    pixels = img_rgb.reshape((-1, 3))
+    kmeans = KMeans(n_clusters=1, random_state=42).fit(pixels)
+    dominant_color = kmeans.cluster_centers_[0].astype(int)
+
+    return image, f"Dominant skin tone RGB: {tuple(dominant_color)}"
+
+# Gradio Interface
 demo = gr.Interface(
     fn=analyze_face,
-    inputs=gr.Image(type="numpy", image_mode="BGR", label="Upload or Capture Face"),
+    inputs=gr.Image(type="numpy", image_mode="BGR", label="Capture or Upload Your Face"),
     outputs=[
-        gr.Image(label="Original Image"),
-        gr.Textbox(label="Detected Face Shape"),
-        gr.Textbox(label="Dominant Skin Tone")
+        gr.Image(type="numpy", label="Face Analysis Output"),
+        gr.Textbox(label="Detected Skin Tone (RGB)")
     ],
-    live=False,
-    allow_flagging="never",
-    title="Step 2: Face Shape and Skin Tone Analyzer"
+    title="Face Scanner for Mask Recommendation",
+    description="Upload or capture a photo to analyze face landmarks and detect skin tone."
 )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()