Skip to content

Latest commit

 

History

History
23 lines (20 loc) · 835 Bytes

run.md

File metadata and controls

23 lines (20 loc) · 835 Bytes
Raw

Training

To start training, run dist_train with DSPDet3D configs.
For ScanNet-md40:

bash tools/dist_train.sh configs/dspdet3d/dspdet3d_scannet-3d-22class.py 2 --work-dir work_dirs/scannet_md40

For TO-SCENE-down:

bash tools/dist_train.sh configs/dspdet3d/dspdet3d_toscene-3d-70class.py 2 --work-dir work_dirs/toscene_down

Testing

Test pre-trained model using dist_test with DSPDet3D configs. We also provide checkpoints used in the paper for easy reproduction.
For ScanNet-md40:

bash tools/dist_test.sh configs/dspdet3d/dspdet3d_scannet-3d-22class.py work_dirs/scannet_md40/latest.pth 2

For TO-SCENE-down:

bash tools/dist_test.sh configs/dspdet3d/dspdet3d_toscene-3d-70class.py work_dirs/toscene_down/latest.pth 2