Shengxi Li
Shengxi Li
[help]How to export swin model to ONNX? Problem: Node (Concat_246) Op (Concat) [ShapeInferenceError]
change "attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0)" to "attn = attn.view(-1, self.num_heads, N, N) + mask.unsqueeze(1)" can solve the problem.
[help]How to export swin model to ONNX? Problem: Node (Concat_246) Op (Concat) [ShapeInferenceError]
> only change this line? yes
Great, i'm looking forward to your paper and can you remind me when paper is completed? My wechat name is xizi_fish and we can add as a friend.
origin VGG-SSD 有38*38的feature map并用作检测，但MobileNet-SSD里面没有把38*38的feature map用于检测。
You can solve the processing time problem from https://github.com/yonghenglh6/DepthwiseConvolution
I also encount this problem.
There are some papers reference to feature normalize such as "normface" or "L2-constrained softmax loss".