About the problem of missing model weight files for training

[chinagalaxy2002]

When I run CUDA_VISIBLE_DEVICES=0 python tools/train.py configs/ESAM/ESAM_online_scannet.py --work-dir work_dirs/ESAM_online_scannet/
Error:
FileNotFoundError: work_dirs/ESAM_sv_scannet/epoch_128.pth can not be found.

But I looked for the document you provided and did not find the model file. Can the author provide a copy?

[xuxw98]

We provide checkpoint for evaluation on online setting. If you want to train ESAM, you should first train a SV model and then train online model, rather than directly training online model. Please follow our README for more details.

[Chase-Arthur]

We provide checkpoint for evaluation on online setting. If you want to train ESAM, you should first train a SV model and then train online model, rather than directly training online model. Please follow our README for more details.

Thank you for your excellent work!
I would like to ask if the weights provided by your github project should be online, right? Is there SV weights that can be provided?

[xuxw98]

Yes. We only provide the final online checkpoint. We are sorry that we have not stored the intermediate checkpoints.
Note that it only takes about 1 day to train SV model on a single RTX 3090 GPU, which is faster than online training.