RuntimeError: The size of tensor a (13) must match the size of tensor b (70135) at non-singleton dimension 3 about fewshot_detection

[SoonFa]

pred_boxes[0] = x.data.view(nB*nA*nH*nW) + grid_x
pred_boxes[1] = y.data.view(nB*nA*nH*nW) + grid_y
pred_boxes[2] = torch.exp(w.data).view(nB*nA*nH*nW) * anchor_w
pred_boxes[3] = torch.exp(h.data).view(nB*nA*nH*nW) * anchor_h

[risemeup]

@bahulkark hello,I met the same mistake with you. Have you fixed it ?

[bahulkark]

Nope.

…

On Mon, 7 Sep 2020, 17:37 risemeup, ***@***.***> wrote: @bahulkark <https://github.com/bahulkark> hello,I met the same mistake with you. Have you fixed it ? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#42 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ALYEWBGGKQS2DIG2AWPV2D3SETEINANCNFSM4OPT4E5A> .

[tailororrr]

I met the same mistake with you

[azharshaikh06]

Anyone have a fix for this?

[cgg99]

Is the pytorch version wrong?

[azharshaikh06]

I am using torch 1.4 and can perform the forward pass through the model error is in the regionloss.py pred_boxes[0] = x.data + grid_x here x.data is a 4D tensor while grid_x is a 2D tensor

[zhanghao5]

I am using pytorch 1.7. With the view() method I can reshape and fit in the addition
pred_boxes[0] = x.data.view(nBnAnHnW) + grid_x
pred_boxes[1] = y.data.view(nBnAnHnW)+ grid_y
pred_boxes[2] = torch.exp(w.data).view(nBnAnHnW) * anchor_w
pred_boxes[3] = torch.exp(h.data).view(nBnAnHnW) * anchor_h

[HeuristicLU]

change region_loss.py
294-297:
pred_boxes[0] = x.data.view(nBnAnHnW) + grid_x
pred_boxes[1] = y.data.view(nBnAnHnW) + grid_y
pred_boxes[2] = torch.exp(w.data).view(nBnAnHnW) * anchor_w
pred_boxes[3] = torch.exp(h.data).view(nBnAnHnW) * anchor_h

and this problem can be resolved