Holmeyoung crnn-pytorch issues

Holmeyoung / crnn-pytorch

Pytorch implementation of CRNN (CNN + RNN + CTCLoss) for all language OCR.

MIT License

378 stars 105 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

size mismatch for rnn.1.embedding.weight

#75 zexuyann closed 7 months ago
0
val gpu slow

#74 Zrufy closed 2 years ago
1
[Friendly reminder] About the accuracy of demo.py

#73 HaliComing closed 2 years ago
0
val loss：nan, accuray：0

#72 Yumisilence opened 2 years ago
1
运行demo用cenn.pth预训练模型显示Expected 512, got 64

#71 whistbibi opened 3 years ago
0
img.sub_(0.5).div_(0.5)

#70 Zrufy opened 3 years ago
0
梯度爆炸,loss显示持续显示为inf

#69 RobertXiaobinW opened 3 years ago
1
Training Problems

#68 zhengyussss closed 3 years ago
2
number images train

#67 Zrufy opened 3 years ago
0
Traanning Question

#66 DLWangSan opened 3 years ago
3
create image Tensor

#65 Hknight514880265 opened 4 years ago
2
No predictions when training or testing the net !!!

#64 hualitlc opened 4 years ago
0
Mean vals and norm vals

#63 Zrufy closed 4 years ago
0
Problem about ctc_loss variable input_length while training

#62 3a532028 closed 4 years ago
1
A question about pre-trained model

#61 dy1998 opened 4 years ago
3
Train problem

#60 susanin1970 opened 4 years ago
1
Output indicates "PAD" char for all columns

#59 SriramPingali opened 4 years ago
3
inference accuracy

#58 Zrufy closed 4 years ago
10
Problem loading checkpointed model

#57 NavneetSajwan closed 4 years ago
0
KeyError : ' '

#56 livekeys opened 4 years ago
1
Batchnormalization layer

#55 Zrufy closed 4 years ago
0
Log_softmax issue

#54 Zrufy closed 4 years ago
4
Conversion to Ncnn

#53 Zrufy closed 4 years ago
0
issue of keep ratio = True

#52 ghost opened 4 years ago
4
数据集尺寸问题

#51 VixeruntR closed 4 years ago
6
maybe the code itself support training with text length > 26

#50 ghost opened 4 years ago
6
Dataset takes up only 1000 images

#49 NavneetSajwan closed 4 years ago
5
how to recognize blank, recognize English and Chinese in one model

#48 ghost opened 4 years ago
8
why set dealwith_lossnan = False by default

#47 ghost closed 4 years ago
3
why crnn.zero_grad() was removed in training?

#46 ghost closed 4 years ago
1
How to create train and val data?

#45 NavneetSajwan closed 4 years ago
4
The loss was very low at the beginning

#44 lim-0 opened 4 years ago
2
training with variable length images and text

#43 ghost closed 4 years ago
2
測試資料不同長度圖片,但預測字數都是固定

#42 ekils closed 4 years ago
3
crnn acc =0

#41 ekils opened 4 years ago
5
I tried to reproduce your code, but failed, cost stuck at 7.0

#40 jaysimon closed 4 years ago
3
Incorrect number of classes

#39 maykulkarni opened 4 years ago
4
Zero accuracy

#38 firesans opened 4 years ago
1
assertion error arise

#37 monica061 closed 4 years ago
2
why use Variable to transform Tensor

#36 Loovelj closed 4 years ago
0
训练loss一开始就是0.053正常么

#35 DuckJ opened 4 years ago
0
CTC 在训练几个 iter 后输出全为 0 的问题

#34 BriFuture closed 5 years ago
6
add Google drive link to download pretrained model

#33 mineshmathew closed 5 years ago
0
When loading pretrained model with mutligpu True, I get the follwoing error

#32 mariembenslama closed 5 years ago
1
When I activate 4 gpus at the same time, I get the following error

#31 mariembenslama opened 5 years ago
29
Fine-tuning

#30 niddal-imam closed 5 years ago
6
Can we merge two checkpoints?

#29 mariembenslama closed 5 years ago
9
字典里面字符的排列顺序重要吗？

#28 deep-practice closed 5 years ago
10
lmdb.Error when create dataset

#27 deep-practice closed 5 years ago
8
Accuracy and epochs

#26 mariembenslama closed 5 years ago
24