Update app.py

app.py

@@ -1,47 +1,55 @@
-import gradio as gr
-import torch
-from transformers import LayoutLMv3ForTokenClassification, LayoutLMv3Processor
+# Install necessary libraries
+!pip install easyocr opencv-python gradio
+
+# Import required libraries
+import cv2
 import easyocr
-import os
-
-# Set a persistent cache directory for EasyOCR
-os.environ['EASYOCR_HOME'] = '/root/.easyocr'
-
-# Pre-download the EasyOCR detection model
-print("Downloading EasyOCR detection model...")
-reader = easyocr.Reader(['en'], download_enabled=True)
-print("EasyOCR detection model downloaded successfully.")
-
-# Load the LayoutLMv3 model and processor
-print("Loading LayoutLMv3 model...")
-model = LayoutLMv3ForTokenClassification.from_pretrained("jordyvl/EElayoutlmv3_jordyvl_rvl_cdip_easyocr_2023-07-09_weighted")
-processor = LayoutLMv3Processor.from_pretrained("jordyvl/EElayoutlmv3_jordyvl_rvl_cdip_easyocr_2023-07-09_weighted")
-print("LayoutLMv3 model loaded successfully.")
-
-def process_image(image):
-    # Perform OCR on the input image
-    print("Performing OCR...")
-    ocr_result = reader.readtext(image)
-    print("OCR completed. Results:", ocr_result)
-
-    # Process the image and OCR result with LayoutLMv3
-    print("Processing image with LayoutLMv3...")
-    inputs = processor(image, ocr_result, return_tensors="pt")
-    outputs = model(**inputs)
-    logits = outputs.logits
-
-    # Return logits for debugging or further processing
-    print("Processing completed. Returning results...")
-    return logits.detach().cpu().numpy()
+import numpy as np
+import gradio as gr
+
+# Function to process the uploaded image and extract text
+def extract_text_from_image(image):
+    # Save the uploaded image to disk
+    image_path = "uploaded_image.jpg"
+    cv2.imwrite(image_path, image)
+
+    # Read the image with OpenCV
+    img = cv2.imread(image_path)
+
+    # Initialize the EasyOCR reader
+    reader = easyocr.Reader(['en', 'ar'], gpu=False)
+
+    # Perform text detection
+    results = reader.readtext(image_path)
+
+    # Draw bounding boxes and overlay text on the image
+    conf_threshold = 0.2
+    for (bbox, text, conf) in results:
+        if conf > conf_threshold:
+            # Get coordinates
+            top_left = tuple(map(int, bbox[0]))
+            bottom_right = tuple(map(int, bbox[2]))
+
+            # Draw rectangle and text
+            img = cv2.rectangle(img, top_left, bottom_right, (0, 0, 255), 2)
+            img = cv2.putText(img, text, top_left, cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 0, 0), 2)
+
+    # Convert the image to RGB (Gradio requires RGB format)
+    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+
+    return img, results
 
 # Define the Gradio interface
 interface = gr.Interface(
-    fn=process_image,
-    inputs=gr.Image(type="pil"),
-    outputs=gr.Textbox(label="Model Output"),
-    live=True  # Enable live updates for faster debugging
+    fn=extract_text_from_image,
+    inputs=gr.Image(type="numpy", label="Upload Image"),
+    outputs=[
+        gr.Image(type="numpy", label="Processed Image"),
+        gr.Text(label="Extracted Text")
+    ],
+    title="Image Text Extractor",
+    description="Upload an image to extract text using EasyOCR.",
 )
 
 # Launch the Gradio app
-print("Launching the Gradio interface...")
 interface.launch()