OPENPCDET_USAGE.md

General OpenPCDet Train/Test usage

Here are just some basic commands for using OpenPCDet.

Training a pretrained model for MS3D

If you'd like to train a separate detector to the ones we've provided, you can use the same config files/commands as OpenPCDet, just make sure to include the DATA_CONFIG._BASE_CONFIG_ as our provided nuscenes_dataset_da.yaml or waymo_dataset_da.yaml. Refer to nuScenes VoxelRCNN (Anchor) as an example.

Test and evaluate the pretrained models

• Test with a pretrained model:

python test.py --cfg_file ${CONFIG_FILE} --batch_size ${BATCH_SIZE} --ckpt ${CKPT}

• To test all the saved checkpoints of a specific training setting and draw the performance curve on the Tensorboard, add the --eval_all argument:

python test.py --cfg_file ${CONFIG_FILE} --batch_size ${BATCH_SIZE} --eval_all

• To test with multiple GPUs:

python -m torch.distributed.launch --nproc_per_node=4 --master_port 47771 test.py --launcher pytorch \
    --cfg_file ${CONFIG_FILE} --batch_size ${BATCH_SIZE}

# or    

sh scripts/dist_test.sh ${NUM_GPUS} \
    --cfg_file ${CONFIG_FILE} --batch_size ${BATCH_SIZE}

# or

sh scripts/slurm_test_mgpu.sh ${PARTITION} ${NUM_GPUS} \
    --cfg_file ${CONFIG_FILE} --batch_size ${BATCH_SIZE}

Train a model

You could optionally add extra command line parameters --batch_size ${BATCH_SIZE}, --epochs ${EPOCHS}, --extra_tag ${EXPERIMENT_NAME} to specify your preferred parameters.

• Train with multiple GPUs or multiple machines

python -m torch.distributed.launch --nproc_per_node=4 --master_port 47771 train.py --launcher pytorch \
    --cfg_file ${CONFIG_FILE} --extra_tag ${EXPERIMENT_NAME}

# or    

sh scripts/dist_train.sh ${NUM_GPUS} --cfg_file ${CONFIG_FILE}

# or 

sh scripts/slurm_train.sh ${PARTITION} ${JOB_NAME} ${NUM_GPUS} --cfg_file ${CONFIG_FILE}

• Train with a single GPU:

python train.py --cfg_file ${CONFIG_FILE}