fix bugs

app.py

@@ -6,25 +6,24 @@ import pytesseract
 from gtts import gTTS
 import io
 
-def preprocess(frame):
-    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
+def preprocess(image):
+    img = np.array(image)
+    img = cv2.flip(img, 1)
+    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
     kernel = np.array([[0,-1,0], [-1,5,-1], [0,-1,0]])
     gray = cv2.filter2D(gray, -1, kernel)
     gray = cv2.fastNlMeansDenoising(gray, None, 30, 7, 21)
     thresh = cv2.adaptiveThreshold(
-        gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
-        cv2.THRESH_BINARY, 31, 10
+        gray, 255,
+        cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
+        cv2.THRESH_BINARY,
+        31, 10
     )
     return thresh
 
-def process_video(frame):
-    if frame is None:
-        return None, "No frame captured", None
-
-    frame = cv2.flip(frame, 1)  # mirror
-    processed = preprocess(frame)
+def extract_text_and_speak(image):
+    processed = preprocess(image)
     processed_pil = Image.fromarray(processed)
-
     text = pytesseract.image_to_string(processed, lang="eng").strip()
     if text == "":
         text = "No readable text found."
@@ -35,17 +34,16 @@ def process_video(frame):
         tts.write_to_fp(buffer)
         buffer.seek(0)
         audio_file = buffer
-
     return processed_pil, text, audio_file
 
 with gr.Blocks() as demo:
-    gr.Markdown("## GabAI - Real-Time OCR with Webcam")
+    gr.Markdown("## GabAI - AI Assistive Reading System")
 
     with gr.Row():
-        webcam = gr.Video(
-            sources=["webcam"],   # corrected argument
-            type="numpy",
-            label="Webcam Feed"
+        webcam = gr.Image(
+            type="pil",
+            sources=["webcam"],
+            label="Webcam Input"
         )
         processed_preview = gr.Image(type="pil", label="Processed Preview")
 
@@ -53,7 +51,7 @@ with gr.Blocks() as demo:
     audio_output = gr.Audio(label="Text-to-Speech Output")
 
     webcam.change(
-        fn=process_video,
+        fn=extract_text_and_speak,
         inputs=webcam,
         outputs=[processed_preview, ocr_text, audio_output]
     )