Macqueen Ma
Macqueen Ma
same quesrion . and I'm mad now
> @abhigoku10, @yossibiton, @azuryl, I have modified DCNv2 from this repository to add the CPU functionality. I have submitted a pull request to Charles Shang, but so far there is...
same error ... even I update gcc 4.8-->5.3 error still pip install inplace_abn `nvcc fatal : Value 'c++14' is not defined for option 'std' error: command '/usr/local/cuda/bin/nvcc' failed with exit...
但是如果训练集图片宽度差异过大，有的几个字，有的几十个字，统一成最大的，有可能训练时CTCloss inf 或者 W ./tensorflow/core/util/ctc/ctc_loss_calculator.h:499] No valid path found. 报错 需要注意
DCN有大神增加了CPU模式,可以用[https://github.com/CharlesShang/DCNv2](https://github.com/CharlesShang/DCNv2)来替换,这个方法最简答.也可以用普通卷积替换动态卷积,只不过效果就不保证了
same question , and I fond that I change my PascalVOC2007 to PascalVOC0712 and I forget to change "/home/user01/Music/fc/py-R-FCN/data/VOCdevkit2007" to "VOCdeviket0712". now I'm retest again. (we needn't to retrain)
which model . I can email you in gmail.
give me your email
I send 5 models in this project in my google drive: https://drive.google.com/drive/folders/1yzrKR-Eu6t4dom4ZyFPFwmmSJzu4v3bb?usp=sharing
The validation set is the same distribution of the training set.（with Simulation by myself） The test set is the same distribution of the real data in the training set （without...