Release code of RandLA-Net with CBL

[whuhxb]

Hi,

Could you please release the code of RandLA-Net with CBL? Thanks.

Best.

[LiyaoTang]

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.

[whuhxb]

@LiyaoTang
Thanks for your kind reply. I will have a try and wait for your code version.

[whuhxb]

@LiyaoTang
Hi, have you released the code of RandLA-Net with CBL? Thanks a lot.
Best.

[MrAirplaneLu]

@LiyaoTang
Hi, have you released the code of RandLA-Net with CBL?

[jwm1d]

Hi, have you released the code of RandLA-Net with CBL? Thanks a lot.
Best.

[MageeChan]

Hi, have you released the code of RandLA-Net with CBL? Thanks a lot.
Best.

[LiyaoTang]

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 However, it makes our pre-trained model no longer valid and requires additional tunning of CBL, which would not be my priority at this moment.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

[zhou991119]

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.