contrastBoundary
contrastBoundary copied to clipboard
Release code of RandLA-Net with CBL
Hi,
Could you please release the code of RandLA-Net with CBL? Thanks.
Best.
Hi, the code with RandLA-Net is still under preparation. We use the original RandLA-Net repo to produce the result, which is not quite compatible with the current framework. For the sake of overall compactness and readability, we are trying to reproduce the RandLA-Net results under the current framework, which could take some time.
If you are in a hurry, I would suggest borrowing the tensorflow implementation of our CBL to RandLA-Net's code. We would be happy to help if you have any problems.
Best.
@LiyaoTang Thanks for your kind reply. I will have a try and wait for your code version.
@LiyaoTang Hi, have you released the code of RandLA-Net with CBL? Thanks a lot. Best.
@LiyaoTang Hi, have you released the code of RandLA-Net with CBL?
Hi, have you released the code of RandLA-Net with CBL? Thanks a lot. Best.
Hi, have you released the code of RandLA-Net with CBL? Thanks a lot. Best.
Hi,
Sorry for my late response, as I have been busy with another code release of our work at nips'23 here, which contains our re-implementation of RandLA-Net and should be compatible with this repo.
While I have tried to incorporate the RandLA-Net official implementation into this repo, I find it quite hard since the code structure is very different. So, I finally decided to re-implement it using the framework of this repo. That is, we can either combine the RandLA-Net from there to this repo, or the CBL head from this repo to there. However, it makes our pre-trained model no longer valid and requires additional tunning of CBL, which would not be my priority at this moment.
For those who are still interested, I would be more than happy to provide help during the training and further implementation if required.
Best, Liyao
Hi, the code with RandLA-Net is still under preparation. We use the original RandLA-Net repo to produce the result, which is not quite compatible with the current framework. For the sake of overall compactness and readability, we are trying to reproduce the RandLA-Net results under the current framework, which could take some time.
If you are in a hurry, I would suggest borrowing the tensorflow implementation of our CBL to RandLA-Net's code. We would be happy to help if you have any problems.
Best.
Hi, Could you please release the original code of RandLA-Net with CBL? Thanks a lot.