Chuming Lin
Chuming Lin
@Light-- Hi, I rerun the code and got the results: Area Under the AR vs AN curve: 68.37602852441032% Do you try to recompile the proposal generation layer? We did not...
@Light-- > cd custom_op/src > make to recompile our proposal feature generation operation. If the result is still incorrect, you can try to modify the file `custom_op/src/Makefile` as below: ```...
@2Zhaokaijie 你需要在nvidia gpu环境下安装nvida驱动和cuda环境，这样才有nvcc这个编译程序
Please download from WeiYun: https://share.weiyun.com/ym38qXe7
The feature used in our paper is TSN feature pretrained on ActivityNet as same as BSN/BMN.
@AshinCarlXiao @sesse The big problem is the work use the tanh function as activate function in the last layer. It is not reasonable. The flow value will be restricted in...
下载下来的视频左上角标注的是4倍5帧的结果，好像跟作者说的有矛盾。
@yangjiangeyjg, thanks for your interest. Because we need to compile our boundary pooling operation in setup.py, C++ compiler is necessary. According your error information, you should install Visual Studio 15...
@yangjiangeyjg I didn't try to compile AFSD in Windows and I am sorry for your problem because of lacking relevant experience.
https://github.com/TencentYoutuResearch/ActionDetection-AFSD/blob/ed86a0df91e58baa7d78c796ed29cff82b1f3fa6/AFSD/common/gen_denseflow_npy.py#L84 use the denseflow will do the operation