AirBernard/Scene-Text-Detection-with-SPCNET

Issues

• NotFoundError: Tensor name "FPN/Conv/BatchNorm/beta/ExponentialMovingAverage" not found in checkpoint files resnet_v1_50.ckp

  #3 opened 5 years ago by aligoglos
  7

• 模型效果有点奇怪

  #14 opened 5 years ago by oysz2016
  2

• 谢谢大佬，有个问题想问问，这个开源如何对弯曲文本进行训练？

  #5 opened 5 years ago by ZDDEAN
  6

• train problem

  #4 opened 5 years ago by WangJie8682
  6

• 训练出model.ckpt文件来了，请问您有检测h-mean的思路吗，现在还不知如何着手

  #21 opened 5 years ago by Dilidulu523
  0

• IC15的训练结果计算hmean有点差

  #20 opened 5 years ago by TuKJet
  0

• Generator use 10 batches for buffering, this may take a while--too long time

  #19 opened 5 years ago by keithyh
  1

• 训练数据的mask标签问题

  #16 opened 5 years ago by youwh-PIRI
  1

• problem of convergence？

  #18 opened 5 years ago by FredlinT
  0

• 结果如何？

  #6 opened 5 years ago by ccckj
  3

• 请问有在ICDAR2017上训练好的权重吗？

  #7 opened 5 years ago by Marscep
  2

• 请问这代码支持多GPU吗？

  #17 opened 5 years ago by FredlinT
  0

• LookupError: gradient registry has no entry for: BroadcastTo

  #15 opened 5 years ago by CindyHuang23
  1

• Cannot create initializer for non-floating point type

  #13 opened 5 years ago by kewin1807
  0

• 训练的icdar2015，出现ValueError: invalid literal for int() with base 10问题

  #11 opened 5 years ago by weiwei23456789
  6

• ValueError: Dimensions must be equal, but are 32 and 256 for 'P5/TCM_Module/BroadcastTo' (op: 'BroadcastTo') with input shapes: [?,32,32], [4] and with input tensors computed as partial shapes: input[1] = [?,32,32,256].

  #10 opened 5 years ago by brooklyn1900
  1

• 测试 P R F 指标

  #12 opened 5 years ago by weiwei23456789
  0

• Can't training this model

  #9 opened 5 years ago by uname0x96
  0

• inconsistent use of tabs and spaces in indentation

  #8 opened 5 years ago by kewin1807
  0

• License?

  #2 opened 5 years ago by FelixSchwarz
  0

• when would the code be released?

  #1 opened 6 years ago by Jiakui
  6