Can you improve your readme content about rac-net HOT 8 OPEN

Thanks for your interesting in our work. We will update the codes after the acceptance of this paper. Thanks.

from rac-net.

Thank you for your reply.And I hope your code can be trained with multi-GPUs,Because my single gpu memory is not enough.Thank you very much

from rac-net.

@wu-zhonghua Congratulations on your paper being accepted by IJCV! I wonder what the pointnet2 directory in the code is used for? And the code about the part of point transformer can be released?

from rac-net.

We have released the code with the Point Transformer backbone. Feel free to ask me if you have further questions.

from rac-net.

@wu-zhonghua Thanks for your excellent work!

from rac-net.

@wu-zhonghua Hi, when I run the code, It shows that forward of memory, and the value of lds_wolf and lds_cda will both be 0. Is this normal, or is my gpu running out of memory?

from rac-net.

Yes, your GPU running out of memory. It is better if you could use more powerful gpu or reduce the batch size.

from rac-net.

@wu-zhonghua Hi, when I run the code, It shows that forward of memory, and the value of lds_wolf and lds_cda will both be 0. Is this normal, or is my gpu running out of memory?

@Yellowshuohahaha I had the same problem as you and was wondering how you ended up solving it? I used two 3090 GPUs running together and set the batch size to 2, and I still get this problem.

from rac-net.