Jaxin

batch is 128 in eval.py, but my testing images are are just 500, the result shows total is 512(4*128). Can you tell me where the additional 12 (512-200) are from?...

default model: validation result : 91.8% testing reult： 84.9% bn and inceptionv3 also exist this difference...

How can I solve this problem.

Picture from web cam效果差？

3

每次运行后的结果都不一致，而且感觉是随机的错误。（每次都是三位测试者）

Each pixel only belong to one label? Even on different people in the same image.

Is there anyone convert the .pth models to .pb models?

2

Is there anyone convert the .pth models to .pb models? With some reasons, Pytorch is not in use while Tensorflow is... Help!!

AttributeError: Can't get attribute 'RD_FPNnet' on <module 'model.T_Net' from '/Users/jasonxin/Documents/dlworkspace/Semantic_Human_Matting/model/T_Net.py'>

1

When I run ./test_camera.sh (--without_gpu), this error occurred: `AttributeError: Can't get attribute 'RD_FPNnet' on ` My torch version is 1.0.0.

loss_t grows when training M-net?

1

loss_t grows when training M-net?

if mask is irregular polygon, this model still work well or not?

You just replace non-person pixels with their gray mode?

1

You just replace non-person pixels with their gray mode?