-
Notifications
You must be signed in to change notification settings - Fork 14
Description
First of all, thanks for open-sourcing this amazing work!
I'm really interested in the novel tasks defined in the paper, especially "Functionality segmentation". The baseline methods are also beautifully designed and perform competitively.
In addition, while trying to follow the benchmark, most existing open-sourced methods designed for fine-grained 3D segmentation perform poorly on this task (for example PartSLIP and 3DAPNet, and even some heavy point-cloud LLMs). The case makes the baseline methods more intriguing, but also makes following the benchmark a bit more difficult to start with.
Is there any plan to release the checkpoints or training codes of any of the baseline methods(e.g. Mask3D-F, SoftGroup-F) ? With the baseline methods as reference, it would greatly help with following your work and participating in the challenge.
Thanks again for reading. I'm looking forward to your reply and learning from your insights!