PETR
This is an README for PETR.
Description
Author: @SekiroRong. This is an implementation of PETR.
Usage
Training commands
In MMDet3D's root directory, run the following command to train the model:
python tools/train.py projects/PETR/configs/petr_vovnet_gridmask_p4_800x320.py
Testing commands
In MMDet3D's root directory, run the following command to test the model:
python tools/test.py projects/PETR/configs/petr_vovnet_gridmask_p4_800x320.py ${CHECKPOINT_PATH}
Results
This Result is trained by petr_vovnet_gridmask_p4_800x320.py and use weights as pretrain weight.
| Backbone | Lr schd | Mem (GB) | Inf time (fps) | mAP | NDS | Download |
|---|---|---|---|---|---|---|
| petr_vovnet_gridmask_p4_800x320 | 1x | 7.62 | 18.7 | 38.3 | 43.5 | model | log |