Spaces:
Running
on
T4
Running
on
T4
JingyeChen22
commited on
Commit
•
cc859d1
1
Parent(s):
9de996f
Update util.py
Browse files
util.py
CHANGED
@@ -26,7 +26,7 @@ for index, c in enumerate(alphabet):
|
|
26 |
|
27 |
|
28 |
|
29 |
-
def transform_mask_pil(mask_root):
|
30 |
"""
|
31 |
This function extracts the mask area and text area from the images.
|
32 |
|
@@ -37,13 +37,13 @@ def transform_mask_pil(mask_root):
|
|
37 |
* The white area is the text area
|
38 |
"""
|
39 |
img = np.array(mask_root)
|
40 |
-
img = cv2.resize(img, (
|
41 |
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
|
42 |
ret, binary = cv2.threshold(gray, 250, 255, cv2.THRESH_BINARY) # pixel value is set to 0 or 255 according to the threshold
|
43 |
return 1 - (binary.astype(np.float32) / 255)
|
|
|
44 |
|
45 |
-
|
46 |
-
def transform_mask(mask_root: str):
|
47 |
"""
|
48 |
This function extracts the mask area and text area from the images.
|
49 |
|
@@ -54,7 +54,7 @@ def transform_mask(mask_root: str):
|
|
54 |
* The white area is the text area
|
55 |
"""
|
56 |
img = cv2.imread(mask_root)
|
57 |
-
img = cv2.resize(img, (
|
58 |
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
|
59 |
ret, binary = cv2.threshold(gray, 250, 255, cv2.THRESH_BINARY) # pixel value is set to 0 or 255 according to the threshold
|
60 |
return 1 - (binary.astype(np.float32) / 255)
|
@@ -125,7 +125,45 @@ def filter_segmentation_mask(segmentation_mask: np.array):
|
|
125 |
|
126 |
|
127 |
|
128 |
-
def combine_image(args, sub_output_dir: str, pred_image_list: List, image_pil: Image, character_mask_pil: Image, character_mask_highlight_pil: Image, caption_pil_list: List):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
129 |
"""
|
130 |
This function combines all the outputs and useful inputs together.
|
131 |
|
@@ -143,20 +181,20 @@ def combine_image(args, sub_output_dir: str, pred_image_list: List, image_pil: I
|
|
143 |
if size == 1:
|
144 |
return pred_image_list[0]
|
145 |
elif size == 2:
|
146 |
-
blank = Image.new('RGB', (
|
147 |
blank.paste(pred_image_list[0],(0,0))
|
148 |
-
blank.paste(pred_image_list[1],(
|
149 |
elif size == 3:
|
150 |
-
blank = Image.new('RGB', (
|
151 |
blank.paste(pred_image_list[0],(0,0))
|
152 |
-
blank.paste(pred_image_list[1],(
|
153 |
-
blank.paste(pred_image_list[2],(
|
154 |
elif size == 4:
|
155 |
-
blank = Image.new('RGB', (
|
156 |
blank.paste(pred_image_list[0],(0,0))
|
157 |
-
blank.paste(pred_image_list[1],(
|
158 |
-
blank.paste(pred_image_list[2],(0,
|
159 |
-
blank.paste(pred_image_list[3],(
|
160 |
|
161 |
|
162 |
return blank
|
@@ -303,4 +341,4 @@ def inpainting_merge_image(original_image, mask_image, inpainting_image):
|
|
303 |
table.append(0)
|
304 |
mask_image = mask_image.point(table, "1")
|
305 |
merged_image = Image.composite(inpainting_image, original_image, mask_image)
|
306 |
-
return merged_image
|
|
|
26 |
|
27 |
|
28 |
|
29 |
+
def transform_mask_pil(mask_root, size):
|
30 |
"""
|
31 |
This function extracts the mask area and text area from the images.
|
32 |
|
|
|
37 |
* The white area is the text area
|
38 |
"""
|
39 |
img = np.array(mask_root)
|
40 |
+
img = cv2.resize(img, (size, size), interpolation=cv2.INTER_NEAREST)
|
41 |
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
|
42 |
ret, binary = cv2.threshold(gray, 250, 255, cv2.THRESH_BINARY) # pixel value is set to 0 or 255 according to the threshold
|
43 |
return 1 - (binary.astype(np.float32) / 255)
|
44 |
+
|
45 |
|
46 |
+
def transform_mask(mask_root, size):
|
|
|
47 |
"""
|
48 |
This function extracts the mask area and text area from the images.
|
49 |
|
|
|
54 |
* The white area is the text area
|
55 |
"""
|
56 |
img = cv2.imread(mask_root)
|
57 |
+
img = cv2.resize(img, (size, size), interpolation=cv2.INTER_NEAREST)
|
58 |
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
|
59 |
ret, binary = cv2.threshold(gray, 250, 255, cv2.THRESH_BINARY) # pixel value is set to 0 or 255 according to the threshold
|
60 |
return 1 - (binary.astype(np.float32) / 255)
|
|
|
125 |
|
126 |
|
127 |
|
128 |
+
def combine_image(args, resolution, sub_output_dir: str, pred_image_list: List, image_pil: Image, character_mask_pil: Image, character_mask_highlight_pil: Image, caption_pil_list: List):
|
129 |
+
"""
|
130 |
+
This function combines all the outputs and useful inputs together.
|
131 |
+
|
132 |
+
Args:
|
133 |
+
args (argparse.ArgumentParser): The arguments.
|
134 |
+
pred_image_list (List): List of predicted images.
|
135 |
+
image_pil (Image): The original image.
|
136 |
+
character_mask_pil (Image): The character-level segmentation mask.
|
137 |
+
character_mask_highlight_pil (Image): The character-level segmentation mask highlighting character regions with green color.
|
138 |
+
caption_pil_list (List): List of captions.
|
139 |
+
"""
|
140 |
+
|
141 |
+
|
142 |
+
size = len(pred_image_list)
|
143 |
+
|
144 |
+
if size == 1:
|
145 |
+
return pred_image_list[0]
|
146 |
+
elif size == 2:
|
147 |
+
blank = Image.new('RGB', (resolution*2, resolution), (0,0,0))
|
148 |
+
blank.paste(pred_image_list[0],(0,0))
|
149 |
+
blank.paste(pred_image_list[1],(resolution,0))
|
150 |
+
elif size == 3:
|
151 |
+
blank = Image.new('RGB', (resolution*3, resolution), (0,0,0))
|
152 |
+
blank.paste(pred_image_list[0],(0,0))
|
153 |
+
blank.paste(pred_image_list[1],(resolution,0))
|
154 |
+
blank.paste(pred_image_list[2],(resolution*2,0))
|
155 |
+
elif size == 4:
|
156 |
+
blank = Image.new('RGB', (resolution*2, resolution*2), (0,0,0))
|
157 |
+
blank.paste(pred_image_list[0],(0,0))
|
158 |
+
blank.paste(pred_image_list[1],(resolution,0))
|
159 |
+
blank.paste(pred_image_list[2],(0,resolution))
|
160 |
+
blank.paste(pred_image_list[3],(resolution,resolution))
|
161 |
+
|
162 |
+
|
163 |
+
return blank
|
164 |
+
|
165 |
+
|
166 |
+
def combine_image_gradio(args, size, sub_output_dir: str, pred_image_list: List, image_pil: Image, character_mask_pil: Image, character_mask_highlight_pil: Image, caption_pil_list: List):
|
167 |
"""
|
168 |
This function combines all the outputs and useful inputs together.
|
169 |
|
|
|
181 |
if size == 1:
|
182 |
return pred_image_list[0]
|
183 |
elif size == 2:
|
184 |
+
blank = Image.new('RGB', (size*2, size), (0,0,0))
|
185 |
blank.paste(pred_image_list[0],(0,0))
|
186 |
+
blank.paste(pred_image_list[1],(size,0))
|
187 |
elif size == 3:
|
188 |
+
blank = Image.new('RGB', (size*3, size), (0,0,0))
|
189 |
blank.paste(pred_image_list[0],(0,0))
|
190 |
+
blank.paste(pred_image_list[1],(size,0))
|
191 |
+
blank.paste(pred_image_list[2],(size*2,0))
|
192 |
elif size == 4:
|
193 |
+
blank = Image.new('RGB', (size*2, size*2), (0,0,0))
|
194 |
blank.paste(pred_image_list[0],(0,0))
|
195 |
+
blank.paste(pred_image_list[1],(size,0))
|
196 |
+
blank.paste(pred_image_list[2],(0,size))
|
197 |
+
blank.paste(pred_image_list[3],(size,size))
|
198 |
|
199 |
|
200 |
return blank
|
|
|
341 |
table.append(0)
|
342 |
mask_image = mask_image.point(table, "1")
|
343 |
merged_image = Image.composite(inpainting_image, original_image, mask_image)
|
344 |
+
return merged_image
|