Yu Zheng
Yu Zheng
https://stackoverflow.com/questions/41048819/how-to-restore-a-model-by-filename-in-tensorflow-r12
@swapgit You could use a sliding window. Say, extract the nearby 15 frame-wise features and feed it into the network. Only my point of view.
The scene labels are stored in the .txt file of each scene. You can refer to the download script provided by the ScanNet Official and add `.txt` to the download...
For the last question, I've tried 256-dim for ScanNet which does not affect the final results though.
> `pc_indices` isn't used in the code anyway. I just commented the line. Then how do you run main_eval.py ?
> I have installed TF 1.14. Still, I am getting the same error. Does anyone have any idea if it has been again moved from TF 1.14 onwards? > same...
I'm also working on this dataset. Do you get the results on nuScene validation split? Are there bugs coming up for the operations below rather than the provided trainval&test split?(Of...
And from my point of view, low performance is pretty normal for a new challenging dataset. You could refer to the paper https://arxiv.org/pdf/1903.11027.pdf.
> The nuscenes dataset is much harder than KITTI. you should focus on nusc dist AP rather than kitti AP to match nuscenes paper result. > @yzheng97 supported version is...
@traveller59 Could you please update the result of `all.pp.config: 50 epoch, 1/8 dataset` ? The result in `README.md` misses the barrier category. The result in `NUSCENES-GUIDE.md` is the same with...