shui-tian-ju-shi

可是requirement里面不是说tensorboardx要1.6吗，你这样直接install会不会装的是最新的？ > torch 1.10.2 coda11.3 两个方式解决： 1、pip install tensorboardX 直接升最新 2、参考了新旧两个版本 在PyTorch 1.6版本中set_training变成了select_model_mode_for_export，改一下就可以了。 说白了就是换了个马甲 @ [liang0502](https://github.com/liang0502)

我通过安装包解决了这个问题，打扰了，谢谢! @ [liang0502](https://github.com/liang0502)

torch==1.0.0 torchvision==0.2.1 tensorboardX==1.6 @[GGgrey](https://github.com/GGgrey)

hi, have you solved this problem? I me the same problem.

hi，my version = 3.7. But that may not be a good fit. shui_tian_ju_shi ***@***.***   ------------------ 原始邮件 ------------------ 发件人: "leoxiaobin/deep-high-resolution-net.pytorch" ***@***.***>; 发送时间: 2022年12月28日(星期三) 凌晨4:01 ***@***.***>; ***@***.******@***.***>; 主题: Re: [leoxiaobin/deep-high-resolution-net.pytorch] Pytroch and tensorboardx versions are...

显示编码有问题，上网查只看到说什么文件打开方式有问题，但我也没找到打开文件的命令~~

感谢您的解决方案，代码成功运行了，而且您的这套代码是我跑过的最方便快捷的代码了，对于代码小白的我来说，感激之情难以言表。我就在这里祝大佬今年继续高中了。

And I can't find where to set the testset_path in eval.py. I mean what dataset in your eval.py to test.

I saw in your paper that, in the inference process, each suspicious face detected should be matched in the Reference set, so the reference set should be very large. Because...

I'm not sure -- what jsonPath does,I think it is to record the authenticity of the pictures extract from the video, real or fake ,am I right?