Update app.py

app.py

@@ -15,7 +15,11 @@ trocr_model = VisionEncoderDecoderModel.from_pretrained('microsoft/trocr-base-ha
 
 def recognize_handwritten_text(image):
     try:
-        # Save the uploaded image to a temporary file
+        # Ensure image is a PIL image and convert to a compatible format
+        if not isinstance(image, Image.Image):
+            image = Image.fromarray(np.array(image)).convert("RGB")
+        
+        # Save the uploaded image to a temporary file in JPEG format
         with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as tmp_file:
             image.save(tmp_file.name, format="JPEG")
             tmp_path = tmp_file.name
@@ -36,9 +40,16 @@ def recognize_handwritten_text(image):
         pil_image = Image.fromarray(processed_image)
         texts = []
         
-        # Recognize text in each detected region
+        # Adjust box unpacking based on actual structure
         for box in boxes:
-            x_min, y_min, x_max, y_max = box[0][0], box[0][1], box[2][0], box[2][1]
+            if len(box) >= 4:  # Check if box has at least 4 coordinates
+                x_min, y_min, x_max, y_max = box[0][0], box[0][1], box[2][0], box[2][1]
+            elif len(box) == 2:  # Handle case with only 2 points (e.g., center and size)
+                x_min, y_min = box[0][0] - box[1][0] / 2, box[0][1] - box[1][1] / 2
+                x_max, y_max = box[0][0] + box[1][0] / 2, box[0][1] + box[1][1] / 2
+            else:
+                continue  # Skip invalid boxes
+            
             crop = pil_image.crop((x_min, y_min, x_max, y_max))
             pixel_values = processor(images=crop, return_tensors="pt").pixel_values
             generated_ids = trocr_model.generate(pixel_values)
@@ -63,10 +74,10 @@ def recognize_handwritten_text(image):
 # Create Gradio interface
 interface = gr.Interface(
     fn=recognize_handwritten_text,
-    inputs=gr.Image(type="pil"),
-    outputs=[gr.Image(type="pil"), gr.Text()],
+    inputs=gr.Image(type="pil", label="Upload any image format"),
+    outputs=[gr.Image(type="pil", label="Detected Text Image"), gr.Text(label="Recognized Text")],
     title="Handwritten Text Detection and Recognition",
-    description="Upload an image to detect and recognize handwritten text."
+    description="Upload an image in any format (JPEG, PNG, BMP, etc.) to detect and recognize handwritten text."
 )
 
 # Launch the app