Update app.py

app.py

@@ -1,37 +1,35 @@
 import gradio as gr
 import torch
 import pytesseract
+import cv2
+import tempfile
 from transformers import AutoTokenizer, AutoModel
 
-# Set Tesseract executable path
-pytesseract.pytesseract.tesseract_cmd = r'/opt/homebrew/bin/tesseract'
+pytesseract.pytesseract.tesseract_cmd = r'/opt/homebrew/bin/tesseract'  # Update this if necessary
 
 # Load the tokenizer and model
 tokenizer_eng = AutoTokenizer.from_pretrained('ucaslcl/GOT-OCR2_0', trust_remote_code=True)
 model_eng = AutoModel.from_pretrained('ucaslcl/GOT-OCR2_0', trust_remote_code=True).eval()
 
 def perform_ocr(image, language):
-    # Convert the Gradio image input to the format suitable for pytesseract
-    img_cv = image  # Assuming image is already in the correct format
-
-    if language == "English":
-        # Perform OCR using the model for English
-        res_eng = model_eng.chat(tokenizer_eng, img_cv, ocr_type='ocr')
-        return res_eng  # Return results for English
-    elif language == "Hindi":
-        # Perform OCR using pytesseract for Hindi
-        res_hin = pytesseract.image_to_string(img_cv, lang='hin', config='--psm 6')
-        return res_hin  # Return results for Hindi
-    else:
-        return "Unsupported language selected."
+    # Save the NumPy array as an image file temporarily
+    with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as temp_file:
+        temp_filename = temp_file.name
+        cv2.imwrite(temp_filename, image)
+
+    # Perform OCR for English
+    res_eng = model_eng.chat(tokenizer_eng, temp_filename, ocr_type='ocr')
+
+    # Clean up temporary file if needed
+    # os.remove(temp_filename)
+
+    return res_eng  # Return results for English
 
 def ocr_and_search(image, language):
     # Call the perform_ocr function
-    extracted_text = perform_ocr(image, language)
-    # You may also want to implement any searching functionality here
-    # ...
+    english_text = perform_ocr(image, language)
     
-    return extracted_text  # Return the OCR result for the selected language
+    return english_text  # Return the OCR result for English
 
 # Create Gradio interface
 iface = gr.Interface(