Update app.py

app.py

@@ -14,29 +14,29 @@ logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(
 
 # Configure Tesseract path (ensure it’s correctly set to your Tesseract installation)
 try:
-    pytesseract.pytesseract.tesseract_cmd = '/usr/bin/tesseract'  # Change path if necessary
-    pytesseract.get_tesseract_version()  # Confirm Tesseract is properly set
-    logging.info("Tesseract is configured properly.")
+    pytesseract.pytesseract.tesseract_cmd = '/usr/bin/tesseract'  # Adjust path if necessary
+    pytesseract.get_tesseract_version()  # Test Tesseract installation
+    logging.info("Tesseract is properly configured.")
 except Exception as e:
     logging.error(f"Tesseract not found or misconfigured: {str(e)}")
 
-# Improved Image Preprocessing function
+# Improved Image Preprocessing function for OCR
 def preprocess_image(img_cv):
     """Enhance the image to improve OCR performance."""
     try:
         # Convert to grayscale
         gray = cv2.cvtColor(img_cv, cv2.COLOR_BGR2GRAY)
-        
-        # Increase contrast
+
+        # Increase contrast using histogram equalization
         contrast = cv2.equalizeHist(gray)
 
         # Apply Gaussian blur to reduce noise
         blurred = cv2.GaussianBlur(contrast, (5, 5), 0)
-        
-        # Adaptive thresholding for binarization
+
+        # Apply adaptive thresholding to binarize the image
         thresh = cv2.adaptiveThreshold(blurred, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, 11, 2)
 
-        # Sharpening the image
+        # Sharpening the image to bring out more details
         sharpened = cv2.filter2D(thresh, -1, np.array([[0, -1, 0], [-1, 5, -1], [0, -1, 0]]))
         
         return sharpened
@@ -44,7 +44,7 @@ def preprocess_image(img_cv):
         logging.error(f"Image preprocessing failed: {str(e)}")
         return img_cv
 
-# Function to extract weight using OCR
+# Function to extract weight from image using OCR
 def extract_weight(img):
     """Extract weight using Tesseract OCR, focusing on digits and decimals."""
     try:
@@ -52,26 +52,31 @@ def extract_weight(img):
             logging.error("No image provided for OCR")
             return "Not detected", 0.0, None
         
+        # Convert the PIL image to OpenCV format
         img_cv = cv2.cvtColor(np.array(img), cv2.COLOR_RGB2BGR)
+
+        # Preprocess the image for better OCR results
         processed_img = preprocess_image(img_cv)
 
-        # Show processed image for debugging
+        # Debug: Show the processed image to verify preprocessing
         debug_img = Image.fromarray(processed_img)
-        debug_img.show()
+        debug_img.show()  # This will open the processed image for debugging purposes
 
-        # Tesseract configuration to extract digits and decimals
+        # Configure Tesseract to detect only digits and decimals
         custom_config = r'--oem 3 --psm 6 -c tessedit_char_whitelist=0123456789.'
-        
+
+        # Use Tesseract OCR to extract text
         text = pytesseract.image_to_string(processed_img, config=custom_config)
         logging.info(f"OCR result: '{text}'")
         
+        # Extract the weight (numbers and decimal)
         weight = ''.join(filter(lambda x: x in '0123456789.', text.strip()))
         
         if weight:
             try:
                 weight_float = float(weight)
                 if weight_float >= 0:
-                    confidence = 95.0  # High confidence
+                    confidence = 95.0  # Assume high confidence if we detect a valid weight
                     logging.info(f"Weight detected: {weight} (Confidence: {confidence:.2f}%)")
                     return weight, confidence, processed_img
             except ValueError:
@@ -85,28 +90,29 @@ def extract_weight(img):
 
 # Main function to process uploaded image and display results
 def process_image(img):
-    """Process the uploaded image, extract weight, and display results."""
+    """Process the uploaded image, extract weight, and return results."""
     if img is None:
         logging.error("No image uploaded")
         return "No image uploaded", None, gr.update(visible=False), gr.update(visible=False)
     
-    # Get the timestamp for IST (Indian Standard Time)
+    # Get timestamp for IST (Indian Standard Time)
     ist_time = datetime.now(pytz.timezone("Asia/Kolkata")).strftime("%d-%m-%Y %I:%M:%S %p")
     
+    # Call the function to extract weight and confidence
     weight, confidence, processed_img = extract_weight(img)
     
-    # If detection failed
+    # If OCR fails to detect weight
     if weight == "Not detected" or confidence < 95.0:
         logging.warning(f"Weight detection failed: {weight} (Confidence: {confidence:.2f}%)")
         return f"{weight} (Confidence: {confidence:.2f}%)", ist_time, gr.update(visible=True), gr.update(visible=False)
     
-    # Convert processed image to base64 for displaying in Gradio
+    # Convert the processed image to base64 format for displaying
     pil_image = Image.fromarray(processed_img)
     buffered = io.BytesIO()
     pil_image.save(buffered, format="PNG")
     img_base64 = base64.b64encode(buffered.getvalue()).decode()
 
-    # Return the detected weight and processed image for Gradio
+    # Return the detected weight, timestamp, and base64 image for Gradio
     return f"{weight} kg (Confidence: {confidence:.2f}%)", ist_time, img_base64, gr.update(visible=True)
 
 # Gradio Interface Setup for Hugging Face