Update app.py

app.py

@@ -8,7 +8,6 @@ import json
 import time
 from gtts import gTTS
 import tempfile
-from PIL import Image
 import mediapipe as mp
 from mediapipe import solutions
 
@@ -19,7 +18,6 @@ with open("gestures_rules.json", "r") as f:
 # ---------------- MEDIAPIPE (CPU ONLY) ----------------
 mp_hands = solutions.hands
 hands = mp_hands.Hands(
-    static_image_mode=False,
     max_num_hands=1,
     min_detection_confidence=0.7,
     min_tracking_confidence=0.7
@@ -36,9 +34,9 @@ def get_finger_states(hand_landmarks):
     ]
 
 def detect_gesture(states):
-    for k, v in gesture_data.items():
-        if v["pattern"] == states:
-            return k
+    for name, rule in gesture_data.items():
+        if rule["pattern"] == states:
+            return name
     return None
 
 def speak_text(text):
@@ -47,8 +45,8 @@ def speak_text(text):
     tts.save(f.name)
     return f.name
 
-# ---------------- VIDEO PROCESS ----------------
-def process_video(frame, sentence, last_char, last_time):
+# ---------------- FRAME PROCESS ----------------
+def process_frame(frame, sentence, last_char, last_time):
     if frame is None:
         return frame, sentence, last_char, last_time
 
@@ -85,14 +83,14 @@ with gr.Blocks(title="Hand2Voice") as demo:
     gr.Markdown("## 🤟 Hand2Voice – Gesture to Speech")
 
     with gr.Row():
-        with gr.Column(scale=1):
-            webcam = gr.Video(
-                sources=["webcam"],
-                streaming=True,
-                height=360
+        with gr.Column():
+            webcam = gr.Image(
+                label="Webcam",
+                type="numpy",
+                live=True
             )
 
-        with gr.Column(scale=1):
+        with gr.Column():
             output = gr.HTML("<h3>Waiting for gestures...</h3>")
             speak_btn = gr.Button("🔊 Speak")
             clear_btn = gr.Button("🧹 Clear")
@@ -102,8 +100,8 @@ with gr.Blocks(title="Hand2Voice") as demo:
     last_char = gr.State("")
     last_time = gr.State(0.0)
 
-    webcam.stream(
-        process_video,
+    webcam.change(
+        process_frame,
         inputs=[webcam, sentence, last_char, last_time],
         outputs=[webcam, sentence, last_char, last_time]
     ).then(