Spaces:
Running
Running
Upload app.py
Browse files
app.py
CHANGED
@@ -39,20 +39,35 @@ def sample_border_color(image, box, padding=2):
|
|
39 |
median_color = np.median(border_pixels, axis=0)
|
40 |
return tuple(map(int, median_color))
|
41 |
|
42 |
-
def detect_text_boxes(image):
|
43 |
-
|
44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
return []
|
46 |
-
|
47 |
-
for line in results[0]:
|
48 |
-
box, (text, confidence) = line
|
49 |
-
if text.strip():
|
50 |
-
x_min = int(min(pt[0] for pt in box))
|
51 |
-
x_max = int(max(pt[0] for pt in box))
|
52 |
-
y_min = int(min(pt[1] for pt in box))
|
53 |
-
y_max = int(max(pt[1] for pt in box))
|
54 |
-
boxes.append(((x_min, y_min, x_max, y_max), text, confidence))
|
55 |
-
return boxes
|
56 |
|
57 |
def remove_text_dynamic_fill(img_path, output_path):
|
58 |
image = cv2.imread(img_path)
|
@@ -106,8 +121,23 @@ def process_folder(input_files):
|
|
106 |
output_path = os.path.join(temp_output, filename)
|
107 |
remove_text_dynamic_fill(file.name, output_path)
|
108 |
|
109 |
-
zip_path =
|
|
|
110 |
return zip_path
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
111 |
|
112 |
demo = gr.Interface(
|
113 |
fn=process_folder,
|
|
|
39 |
median_color = np.median(border_pixels, axis=0)
|
40 |
return tuple(map(int, median_color))
|
41 |
|
42 |
+
def detect_text_boxes(image, max_dim=1280):
|
43 |
+
try:
|
44 |
+
# Resize large images to reduce memory load
|
45 |
+
height, width = image.shape[:2]
|
46 |
+
if max(height, width) > max_dim:
|
47 |
+
scale = max_dim / float(max(height, width))
|
48 |
+
image = cv2.resize(image, (int(width * scale), int(height * scale)))
|
49 |
+
|
50 |
+
results = PaddleOCR.ocr(image, cls=True)
|
51 |
+
|
52 |
+
if results is None or not results[0]:
|
53 |
+
print("No OCR results found or OCR returned None.")
|
54 |
+
return []
|
55 |
+
|
56 |
+
boxes = []
|
57 |
+
for line in results[0]:
|
58 |
+
box, (text, confidence) = line
|
59 |
+
if text.strip():
|
60 |
+
x_min = int(min(pt[0] for pt in box))
|
61 |
+
x_max = int(max(pt[0] for pt in box))
|
62 |
+
y_min = int(min(pt[1] for pt in box))
|
63 |
+
y_max = int(max(pt[1] for pt in box))
|
64 |
+
boxes.append(((x_min, y_min, x_max, y_max), text, confidence))
|
65 |
+
return boxes
|
66 |
+
|
67 |
+
except Exception as e:
|
68 |
+
print(f"OCR failed on image: {e}")
|
69 |
return []
|
70 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
71 |
|
72 |
def remove_text_dynamic_fill(img_path, output_path):
|
73 |
image = cv2.imread(img_path)
|
|
|
121 |
output_path = os.path.join(temp_output, filename)
|
122 |
remove_text_dynamic_fill(file.name, output_path)
|
123 |
|
124 |
+
zip_path = "/tmp/cleaned_output.zip"
|
125 |
+
zip_folder(output_path, zip_path)
|
126 |
return zip_path
|
127 |
+
|
128 |
+
import zipfile
|
129 |
+
|
130 |
+
def zip_folder(folder_path, output_zip):
|
131 |
+
with zipfile.ZipFile(output_zip, 'w', zipfile.ZIP_DEFLATED) as zipf:
|
132 |
+
for root, _, files in os.walk(folder_path):
|
133 |
+
for file in files:
|
134 |
+
file_path = os.path.join(root, file)
|
135 |
+
arcname = os.path.relpath(file_path, folder_path)
|
136 |
+
zipf.write(file_path, arcname)
|
137 |
+
|
138 |
+
|
139 |
+
|
140 |
+
|
141 |
|
142 |
demo = gr.Interface(
|
143 |
fn=process_folder,
|