Open evbtst opened 4 years ago
line 66: msv_E2[sc] = upsample(F.softmax(e2[0], dim=1)[:,1].data.cpu(), (h,w))
msv_E2[sc] = upsample(F.softmax(e2[0], dim=1)[:,1].data.cpu(), (h,w))
line 71: msv_E2[sc] = F.softmax(e2[0], dim=1)[:,1].data.cpu()
msv_E2[sc] = F.softmax(e2[0], dim=1)[:,1].data.cpu()
If you run these lines for each object (Propagate_MS is inside the loop for o in range(num_objects) ) , why use softmax instead of sigmoid?
for o in range(num_objects)
Hi @evbtst, We perform 2-way (FG & BG) softmax for the binary segmentation. The similar operation can be implemented using the sigmoid. But, we simply choose to use the softmax.
line 66:
msv_E2[sc] = upsample(F.softmax(e2[0], dim=1)[:,1].data.cpu(), (h,w))
line 71:
msv_E2[sc] = F.softmax(e2[0], dim=1)[:,1].data.cpu()
If you run these lines for each object (Propagate_MS is inside the loop
for o in range(num_objects)
) , why use softmax instead of sigmoid?