Update app.py

app.py

@@ -7,17 +7,6 @@ import numpy as np
 import gradio as gr
 from paddleocr import PaddleOCR
 
-from PIL import Image
-
-def is_valid_image(path):
-    try:
-        img = Image.open(path)
-        img.verify()
-        return True
-    except:
-        return False
-
-
 ocr = PaddleOCR(use_angle_cls=True, lang='en', det_model_dir='models/det', rec_model_dir='models/rec', cls_model_dir='models/cls')
 
 def classify_background_color(avg_color, white_thresh=230, black_thresh=50, yellow_thresh=100):
@@ -50,48 +39,20 @@ def sample_border_color(image, box, padding=2):
     median_color = np.median(border_pixels, axis=0)
     return tuple(map(int, median_color))
 
-def detect_text_boxes(image, max_dim=1280):
-    try:
-        # Check if image is valid
-        if image is None or not hasattr(image, 'shape'):
-            print("Invalid image. Skipping...")
-            return []
-
-        # Resize large images to reduce memory load
-        height, width = image.shape[:2]
-        if max(height, width) > max_dim:
-            scale = max_dim / float(max(height, width))
-            image = cv2.resize(image, (int(width * scale), int(height * scale)))
-
-        # Ensure image is in RGB
-        if image.shape[2] == 1:
-            image = cv2.cvtColor(image, cv2.COLOR_GRAY2RGB)
-        elif image.shape[2] == 3:
-            image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
-
-        # Call PaddleOCR correctly
-        results = ocr.ocr(image, cls=True)
-
-        if results is None or not results[0]:
-            print("No OCR results found or OCR returned None.")
-            return []
-
-        boxes = []
-        for line in results[0]:
-            box, (text, confidence) = line
-            if text.strip():
-                x_min = int(min(pt[0] for pt in box))
-                x_max = int(max(pt[0] for pt in box))
-                y_min = int(min(pt[1] for pt in box))
-                y_max = int(max(pt[1] for pt in box))
-                boxes.append(((x_min, y_min, x_max, y_max), text, confidence))
-        return boxes
-
-    except Exception as e:
-        print(f"OCR failed on image: {e}")
+def detect_text_boxes(image):
+    results = ocr.ocr(image, cls=True)
+    if not results or not results[0]:
         return []
-
-
+    boxes = []
+    for line in results[0]:
+        box, (text, confidence) = line
+        if text.strip():
+            x_min = int(min(pt[0] for pt in box))
+            x_max = int(max(pt[0] for pt in box))
+            y_min = int(min(pt[1] for pt in box))
+            y_max = int(max(pt[1] for pt in box))
+            boxes.append(((x_min, y_min, x_max, y_max), text, confidence))
+    return boxes
 
 def remove_text_dynamic_fill(img_path, output_path):
     image = cv2.imread(img_path)
@@ -138,36 +99,16 @@ def remove_text_dynamic_fill(img_path, output_path):
     image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
     cv2.imwrite(output_path, image)
 
-import uuid
-
 def process_folder(input_files):
     temp_output = tempfile.mkdtemp()
-    
     for file in input_files:
         filename = os.path.basename(file.name)
         output_path = os.path.join(temp_output, filename)
         remove_text_dynamic_fill(file.name, output_path)
 
-    unique_name = str(uuid.uuid4())[:8]
-    zip_path = os.path.join("/tmp", f"cleaned_output_{unique_name}.zip")
-    shutil.make_archive(zip_path.replace(".zip", ""), 'zip', temp_output)
-
-    delayed_cleanup(zip_path)
+    zip_path = shutil.make_archive(temp_output, 'zip', temp_output)
     return zip_path
 
-
-import threading
-import time
-
-def delayed_cleanup(path, delay=30):
-    def cleanup():
-        time.sleep(delay)
-        if os.path.exists(path):
-            os.remove(path)
-    threading.Thread(target=cleanup).start()
-
-
-
 demo = gr.Interface(
     fn=process_folder,
     inputs=gr.File(file_types=[".jpg", ".jpeg", ".png"], file_count="multiple", label="Upload Comic Images"),
@@ -176,4 +117,4 @@ demo = gr.Interface(
     description="Upload comic images and get a zip of cleaned versions (text removed). Uses PaddleOCR for detection."
 )
 
-demo.launch()
+demo.launch()