search

HuXiaoling / TopoLoss

Code for the NeurIPS 2019 paper: Topology-Preserving Deep Image Segmentation
MIT License
146 stars 22 forks source link

some question about the batchsize setting #27

Open FZH1802 opened 3 months ago

FZH1802 commented 3 months ago

When I set the batch size to 1, the code works, 微信截图_20240409233251 but when I set the batch size to greater than 1, it will throw an error

HuXiaoling commented 3 months ago

Could you please try computing topoloss on a two-dimensional patch with the size of (H*W) instead of a three- or four-dimensional tensor (B*H*W or B*C*H*W)?

FZH1802 commented 3 months ago

Thank you for your timely response. I am currently using a for loop to solve this problem, but the calculation speed is really too slow. Do you have any better suggestions?

Lenleooo commented 2 months ago

Thank you for your timely response. I am currently using a for loop to solve this problem, but the calculation speed is really too slow. Do you have any better suggestions?

i am also encountered this problem, if possible, we can add the contact information and discuss together.

FZH1802 commented 2 months ago

Thank you for your timely response. I am currently using a for loop to solve this problem, but the calculation speed is really too slow. Do you have any better suggestions?

i am also encountered this problem, if possible, we can add the contact information and discuss together.

好的，可以告诉我你的微信吗