JimReno
JimReno
I copied the code to a python script. And run in local PyCharm with an ssh interpreter from the server. the script works well. Maybe Jupyter notebook has an issue...
@rambleramble I do set the learning rate to 1e5 when I do transfer learning on my own data set. please try to reduce the learning rate
作者都开源了代码、写了详细的部署文档，自己用一用翻译软件吧。感恩开源，感恩贡献！
I have encounterd the same issue. For you installed a newer version of pyyaml. Alternatively, you can modify the `load()` function's parameters: `yaml.load(fo.read(), Loader=yaml.FullLoader)`
Hi Sina: I think color matters. Cause the network was trained using real hand picture. The detect score will be low if you are wearing gloves. You can look into...
添加`center_crop()`会是的validation阶段的结果不稳定。去掉`center_crop()`后的验证结果稳定。 我在fer2013上面做的实验验证的上述说法，至于原因不是很理解，我在validation阶段，去掉了`center_crop()`。