Issues
- 1
[Get text]
#25 opened by cuonghv0298 - 4
input image size
#24 opened by Mchapariniya - 5
demo weights and training times
#18 opened by xen0f0n - 2
- 3
Can I get coordinates of baselines
#15 opened by longwall - 2
- 3
- 19
Train and validation loss are always 0.0000
#9 opened by apirrone - 1
Generating the pixel ground truth images [Converting labeled layout files into ground-truth]
#3 opened - 0
I ran through the code using my own training data. It is text line, X.jpg contains gray image, X_GT0.jpg contains background information. X_GT1.jpg contains baseline information. X_GT2.jpg contains the separator information. For each position the sum is one. But when test the module, all pixels are predicted as separator. Could you supply a sample image so that I can do more analysis?
#22 opened by xinxin664 - 2
demo model fail with low resolution
#21 opened by lynx97 - 1
- 3
- 3
How not to use weight on cbad dataset
#17 opened by elyamanyahmed95 - 1
Train from the demo model
#14 opened by Dragule - 2
Dataset about 2017cBAD test xml groundtruth
#13 opened by lius0-0 - 1
Overfitting issue
#12 opened by apirrone - 3
- 1
- 2
- 1
Ground Truth images
#7 opened by alperdemir03 - 4
train_aru.py path problem
#6 opened by jupinter - 1
No module named 'pix_lab'
#4 opened - 3
Number of training images necessary
#8 opened by apirrone