summerlvsong/Aggregation-Cross-Entropy

Issues

• loss decline but accuracy near to 0

  #25 opened 5 years ago by bjlgcxc
  5

• Need Help! Loss nan

  #2 opened 6 years ago by allen4747
  3

• 是否可以用在中文OCR上面呢

  #27 opened 4 years ago by s1162276945
  2

• Can’t reproduce your results in your cvpr paper

  #17 opened 5 years ago by yflv-yanxia
  3

• It doesn't work on my data, why don't you provide a pretrained model?

  #20 opened 5 years ago by LJXLJXLJX
  5

• Prediction script

  #12 opened 4 years ago
  2

• HWDB results

  #26 opened 4 years ago by DietDietDiet
  0

• Why use ResLSTM for Offline Handwritten Chinese Text Recognition ?

  #13 opened 5 years ago by ivoxx
  1

• Use ACE loss for English handwriting recognition，but not converge

  #14 opened 5 years ago by liudatutu
  2

• combined with CTC?

  #19 opened 5 years ago by liangzimei
  1

• 关于论文公式（2）的理解问题

  #24 opened 5 years ago by yyr6661
  0

• def vis(self,iteration) function, possible error

  #23 opened 5 years ago by shanenian
  0

• What is the accuracy criterion in 2D prediction?

  #22 opened 5 years ago by Cherway
  0

• Would MSE loss work if i remove the softmax?

  #21 opened 5 years ago by Intsigstephon
  0

• I would like to ask you how to accurately predict the character order of a word.

  #8 opened 6 years ago by Meicsu199345
  6

• Compared with CTC, the effect is very poor in Chinese long text

  #18 opened 5 years ago by dlml
  0

• Could you share your trained models?

  #16 opened 5 years ago by dpengwen
  0

• KL Divergence

  #15 opened 5 years ago by viig99
  0

• Do ACE work? same point about HCTR Result

  #11 opened 5 years ago by HannaRiver
  0

• how to get val label

  #9 opened 5 years ago by yangpd
  1

• Would it be helpful in Speech Recognition especially in acoustic model to replace CTC with ACE?

  #10 opened 5 years ago by nemoramo
  1

• can you share 288 cropped natural images for testing in CUTE80

  #7 opened 6 years ago by jiaying96
  1

• code about HCTR will be publicly or not?

  #4 opened 6 years ago by jiaying96
  1

• CUDA_VISIBLE_DEVICES=0 python -u main.py \ 2>&1 | tee $filename

  #6 opened 6 years ago by jiaying96
  2

• could you upload your trained model for scene text recognition?

  #5 opened 6 years ago by lanfeng4659
  1

• Can you provide the data generator tool ?

  #3 opened 6 years ago by lamhoangtung
  4

• When will the code be publicly available?

  #1 opened 6 years ago by allen4747
  0