Spaces:
Build error
Build error
Update visualization
Browse files
app.py
CHANGED
|
@@ -14,6 +14,7 @@ from tempfile import TemporaryDirectory
|
|
| 14 |
import xlsxwriter
|
| 15 |
import matplotlib.pyplot as plt
|
| 16 |
import matplotlib.patches as patches
|
|
|
|
| 17 |
|
| 18 |
from paddleocr import PaddleOCR
|
| 19 |
# import pytesseract
|
|
@@ -218,8 +219,9 @@ def visualize_structure(pil_img, structure_result):
|
|
| 218 |
height = image.shape[0]
|
| 219 |
# print(width, height)
|
| 220 |
|
| 221 |
-
|
| 222 |
-
|
|
|
|
| 223 |
|
| 224 |
for i, result in enumerate(structure_result):
|
| 225 |
class_id = int(result[5])
|
|
@@ -250,47 +252,90 @@ def visualize_structure(pil_img, structure_result):
|
|
| 250 |
# Hatch
|
| 251 |
rect = patches.Rectangle(bbox[:2], bbox[2]-bbox[0], bbox[3]-bbox[1],
|
| 252 |
linewidth=1, alpha=0.4,
|
| 253 |
-
edgecolor=color,facecolor='none',
|
| 254 |
linestyle='--',hatch=hatch)
|
| 255 |
ax.add_patch(rect)
|
| 256 |
# Edge
|
| 257 |
rect = patches.Rectangle(bbox[:2], bbox[2]-bbox[0], bbox[3]-bbox[1],
|
| 258 |
linewidth=linewidth,
|
| 259 |
-
edgecolor=color,facecolor='none',
|
| 260 |
linestyle="--")
|
| 261 |
ax.add_patch(rect)
|
| 262 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 263 |
plt.axis('off')
|
| 264 |
img_buf = io.BytesIO()
|
| 265 |
-
plt.savefig(img_buf, bbox_inches='tight', dpi=
|
|
|
|
| 266 |
|
| 267 |
return PIL.Image.open(img_buf)
|
| 268 |
|
| 269 |
|
| 270 |
def visualize_cells(pil_img, cells):
|
| 271 |
-
|
| 272 |
-
|
|
|
|
| 273 |
|
| 274 |
-
for
|
| 275 |
bbox = cell['bbox']
|
| 276 |
-
|
|
|
|
|
|
|
|
|
|
| 277 |
alpha = 0.3
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 278 |
else:
|
| 279 |
-
|
| 280 |
-
|
| 281 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 282 |
ax.add_patch(rect)
|
| 283 |
-
rect = patches.Rectangle(bbox[:2], bbox[2]-bbox[0], bbox[3]-bbox[1], linewidth=
|
| 284 |
-
|
| 285 |
-
alpha=0.08, hatch='///')
|
| 286 |
ax.add_patch(rect)
|
| 287 |
-
rect = patches.Rectangle(bbox[:2], bbox[2]-bbox[0], bbox[3]-bbox[1], linewidth=
|
| 288 |
-
|
| 289 |
ax.add_patch(rect)
|
| 290 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 291 |
plt.axis('off')
|
| 292 |
img_buf = io.BytesIO()
|
| 293 |
-
plt.savefig(img_buf, bbox_inches='tight', dpi=
|
|
|
|
| 294 |
|
| 295 |
return PIL.Image.open(img_buf)
|
| 296 |
|
|
|
|
| 14 |
import xlsxwriter
|
| 15 |
import matplotlib.pyplot as plt
|
| 16 |
import matplotlib.patches as patches
|
| 17 |
+
from matplotlib.patches import Patch
|
| 18 |
|
| 19 |
from paddleocr import PaddleOCR
|
| 20 |
# import pytesseract
|
|
|
|
| 219 |
height = image.shape[0]
|
| 220 |
# print(width, height)
|
| 221 |
|
| 222 |
+
plt.imshow(pil_img, interpolation='lanczos')
|
| 223 |
+
plt.gcf().set_size_inches(20, 20)
|
| 224 |
+
ax = plt.gca()
|
| 225 |
|
| 226 |
for i, result in enumerate(structure_result):
|
| 227 |
class_id = int(result[5])
|
|
|
|
| 252 |
# Hatch
|
| 253 |
rect = patches.Rectangle(bbox[:2], bbox[2]-bbox[0], bbox[3]-bbox[1],
|
| 254 |
linewidth=1, alpha=0.4,
|
| 255 |
+
edgecolor=color, facecolor='none',
|
| 256 |
linestyle='--',hatch=hatch)
|
| 257 |
ax.add_patch(rect)
|
| 258 |
# Edge
|
| 259 |
rect = patches.Rectangle(bbox[:2], bbox[2]-bbox[0], bbox[3]-bbox[1],
|
| 260 |
linewidth=linewidth,
|
| 261 |
+
edgecolor=color, facecolor='none',
|
| 262 |
linestyle="--")
|
| 263 |
ax.add_patch(rect)
|
| 264 |
|
| 265 |
+
plt.xticks([], [])
|
| 266 |
+
plt.yticks([], [])
|
| 267 |
+
|
| 268 |
+
legend_elements = []
|
| 269 |
+
for class_name in structure_class_names:
|
| 270 |
+
color, alpha, linewidth, hatch = get_bbox_decorations('recognition', structure_class_map[class_name])
|
| 271 |
+
legend_elements.append(
|
| 272 |
+
Patch(facecolor=color, edgecolor=color, label=class_name, hatch=hatch, alpha=alpha)
|
| 273 |
+
)
|
| 274 |
+
|
| 275 |
+
plt.legend(handles=legend_elements, bbox_to_anchor=(0.5, -0.02), loc='upper center', borderaxespad=0,
|
| 276 |
+
fontsize=10, ncol=3)
|
| 277 |
+
plt.gcf().set_size_inches(10, 10)
|
| 278 |
plt.axis('off')
|
| 279 |
img_buf = io.BytesIO()
|
| 280 |
+
plt.savefig(img_buf, bbox_inches='tight', dpi=150)
|
| 281 |
+
plt.close()
|
| 282 |
|
| 283 |
return PIL.Image.open(img_buf)
|
| 284 |
|
| 285 |
|
| 286 |
def visualize_cells(pil_img, cells):
|
| 287 |
+
plt.imshow(pil_img, interpolation="lanczos")
|
| 288 |
+
plt.gcf().set_size_inches(20, 20)
|
| 289 |
+
ax = plt.gca()
|
| 290 |
|
| 291 |
+
for cell in cells:
|
| 292 |
bbox = cell['bbox']
|
| 293 |
+
|
| 294 |
+
if cell['column header']:
|
| 295 |
+
facecolor = (1, 0, 0.45)
|
| 296 |
+
edgecolor = (1, 0, 0.45)
|
| 297 |
alpha = 0.3
|
| 298 |
+
linewidth = 2
|
| 299 |
+
hatch='//////'
|
| 300 |
+
elif cell['projected row header']:
|
| 301 |
+
facecolor = (0.95, 0.6, 0.1)
|
| 302 |
+
edgecolor = (0.95, 0.6, 0.1)
|
| 303 |
+
alpha = 0.3
|
| 304 |
+
linewidth = 2
|
| 305 |
+
hatch='//////'
|
| 306 |
else:
|
| 307 |
+
facecolor = (0.3, 0.74, 0.8)
|
| 308 |
+
edgecolor = (0.3, 0.7, 0.6)
|
| 309 |
+
alpha = 0.3
|
| 310 |
+
linewidth = 2
|
| 311 |
+
hatch='\\\\\\\\\\\\'
|
| 312 |
+
|
| 313 |
+
rect = patches.Rectangle(bbox[:2], bbox[2]-bbox[0], bbox[3]-bbox[1], linewidth=linewidth,
|
| 314 |
+
edgecolor='none',facecolor=facecolor, alpha=0.1)
|
| 315 |
ax.add_patch(rect)
|
| 316 |
+
rect = patches.Rectangle(bbox[:2], bbox[2]-bbox[0], bbox[3]-bbox[1], linewidth=linewidth,
|
| 317 |
+
edgecolor=edgecolor,facecolor='none',linestyle='-', alpha=alpha)
|
|
|
|
| 318 |
ax.add_patch(rect)
|
| 319 |
+
rect = patches.Rectangle(bbox[:2], bbox[2]-bbox[0], bbox[3]-bbox[1], linewidth=0,
|
| 320 |
+
edgecolor=edgecolor,facecolor='none',linestyle='-', hatch=hatch, alpha=0.2)
|
| 321 |
ax.add_patch(rect)
|
| 322 |
|
| 323 |
+
plt.xticks([], [])
|
| 324 |
+
plt.yticks([], [])
|
| 325 |
+
|
| 326 |
+
legend_elements = [Patch(facecolor=(0.3, 0.74, 0.8), edgecolor=(0.3, 0.7, 0.6),
|
| 327 |
+
label='Data cell', hatch='\\\\\\\\\\\\', alpha=0.3),
|
| 328 |
+
Patch(facecolor=(1, 0, 0.45), edgecolor=(1, 0, 0.45),
|
| 329 |
+
label='Column header cell', hatch='//////', alpha=0.3),
|
| 330 |
+
Patch(facecolor=(0.95, 0.6, 0.1), edgecolor=(0.95, 0.6, 0.1),
|
| 331 |
+
label='Projected row header cell', hatch='//////', alpha=0.3)]
|
| 332 |
+
plt.legend(handles=legend_elements, bbox_to_anchor=(0.5, -0.02), loc='upper center', borderaxespad=0,
|
| 333 |
+
fontsize=10, ncol=3)
|
| 334 |
+
plt.gcf().set_size_inches(10, 10)
|
| 335 |
plt.axis('off')
|
| 336 |
img_buf = io.BytesIO()
|
| 337 |
+
plt.savefig(img_buf, bbox_inches='tight', dpi=150)
|
| 338 |
+
plt.close()
|
| 339 |
|
| 340 |
return PIL.Image.open(img_buf)
|
| 341 |
|