Tai Wang
Close due to inactivity. Please feel free to reopen it if there are any new questions.
We will clarify the benchmark settings and update the results in the paper when polishing the benchmarks next. For now, we show the benchmark results on the validation set in...
You can directly refer to the latest results shown in the GitHub repo. We will consider updating the results in the paper during polishing the EmbodiedScan and MMScan benchmarks in...
We provide a mini set to achieve lightweight training. You can follow that benchmark result to reduce the burden of experiments.
We will release the baselines for MMScan in the next weeks. For EmbodiedScan, if you have any interested baseline, we may consider sorting out the code with a higher priority...
OK, LEO is included in our release plan for MMScan's baselines. Please stay tuned and we will try to release them ASAP.
> More discussion in [#696#issuecomment-874275857](https://github.com/open-mmlab/mmdetection3d/pull/696#issuecomment-874275857). > > It's WIP now as we also need a unit test and probably ScanNet + `MultiViewPipeline` test. Can I add 2-3 ScanNet .jpg images...
> Hi, @Tai-Wang > I think the PR is fine. Can you please remove `WIP` and have a look? Great job! There is only something about design details that can...