Update app.py

app.py

@@ -2,20 +2,34 @@ import gradio as gr
 from paddleocr import PaddleOCR
 from PIL import Image, ImageOps
 import numpy as np
+import cv2
+import io
 
 # Initialize PaddleOCR
 ocr = PaddleOCR(use_angle_cls=True, lang='en')
 
-# Function to process uploaded/captured images
-def process_uploaded_image(image, camera_type="rear"):
-    if image is None:
-        return "No image uploaded."
-    
-    # Mirror the image for rear camera
+# Function to extract an image from video and process it
+def process_video_frame(video_data, camera_type="rear"):
+    if video_data is None:
+        return "No video feed provided."
+
+    # Decode video bytes
+    video_stream = io.BytesIO(video_data)
+    video_stream.seek(0)
+    video_array = np.frombuffer(video_stream.read(), np.uint8)
+    video_frame = cv2.imdecode(video_array, cv2.IMREAD_COLOR)
+
+    if video_frame is None:
+        return "Failed to decode video feed."
+
+    # Convert frame to PIL Image
+    image = Image.fromarray(cv2.cvtColor(video_frame, cv2.COLOR_BGR2RGB))
+
+    # Flip the image horizontally for rear camera
     if camera_type == "rear":
         image = ImageOps.mirror(image)
-    
-    # Perform OCR on the corrected image
+
+    # Perform OCR
     result = ocr.ocr(np.array(image))
     extracted_text = []
     for line in result[0]:
@@ -27,9 +41,9 @@ def create_app():
     with gr.Blocks() as app:
         gr.Markdown("## OCR Processor with Rear Camera Fix")
         
-        # Input for image capture and camera type selection
+        # Inputs for video feed and camera type
         with gr.Row():
-            image_input = gr.Image(source="webcam", type="pil", label="Capture Image from Webcam")
+            video_input = gr.Video(label="Capture Video Feed")
             camera_type = gr.Dropdown(
                 choices=["rear", "front"],
                 value="rear",
@@ -39,11 +53,11 @@ def create_app():
         # Output for extracted text
         extracted_text = gr.Textbox(label="Extracted Text")
         
-        # Button to process the image
-        process_button = gr.Button("Process Image")
+        # Button to process the video feed
+        process_button = gr.Button("Process Frame for OCR")
         process_button.click(
-            fn=process_uploaded_image,
-            inputs=[image_input, camera_type],
+            fn=process_video_frame,
+            inputs=[video_input, camera_type],
             outputs=extracted_text
         )