GaitSet icon indicating copy to clipboard operation
GaitSet copied to clipboard

A flexible, effective and fast cross-view gait recognition network

Results 108 GaitSet issues
Sort by recently updated
recently updated
newest added

作者您好,请问test报这个错,该怎么解决呢? @AbnerHqC ValueError: need at least one array to concatenate

训练时一直卡在 Training START ![image](https://user-images.githubusercontent.com/71956241/96574265-ccfd9d00-1301-11eb-94a9-c09ed300f686.png)

请问在FramePipeline中直接将(n,m,d)输入seperate triplet loss,为何是seperate triplet loss,其seperate是每个scale的特征1个triplet loss吗?为何不直接像testing时拉直整个feature为(batch_size, hidden_dim)进而用1个“普通”的triplet loss?

gaitset.py的输出应该是(n,m,d)吧,在model.py中输入triplet_loss前, triplet_feature = feature.permute(1, 0, 2).contiguous() triplet_label = target_label.unsqueeze(0).repeat(triplet_feature.size(0), 1) 请问这样不就将feature变为(m,n,d)了嘛,triplet_loss代码看的不是很懂,注释的输入是(n,m,d). 另外,请问这里边n是否是batch_size,m=62, d=256,不知道理解的对不对?因为论文中HPP部分用n表示了strip是个数即62,其他部分还用n表示了帧数

There were always print an warning before: ```python [W LegacyDefinitions.cpp:55] Warning: masked_scatter_ received a mask with dtype torch.uint8, this behavior is now deprecated,please use a mask with dtype torch.bool instead....

I am new to this ML field, can anyone plz tell me how to run this code. i.e to get the output

您好,我想知道他们是如何判别某个文件夹测试是否正确的,这个识别准确率怎么算出来的,另外是否可以单独输入某个轮廓文件夹,然后识别并输出此人的名称。急等,希望得到大佬的帮助

您好,我想请问一下,您为什么设置margin=0.2,这样效果是最好的吗?如果修改margin=0.5的话,是否triplet的训练不易收敛?谢谢