huangyuan
huangyuan
现在可以训练了，首先要注意中文编码，都使用utf-8编码，然后更改了config文件，把chars替换成自己数据集的中文字符串
我也遇到了这个问题，我发现是我使用的Pytorch版本太高，换回了pytorch1.4，现在编译成功了
The camera intrinsic will change with image crop, scale, and flip, please refer to section 4.1 in the paper and nyu_dataset. About your questions: 1. The nyu_dataset contains two different...
Check environment.yml Try`pip install timm==0.6.12`
> You could try swapping the inputs for the left and right views. It could be that the algorithm is treating the left view as the right view. Thank you...