M_P_C_F
1 UESTC
We present M_P_C_F for point cloud and pseduo point cloud.
This is the official implementation of [MPCF], built on SFD and OpenPCDet
🔥Highlights
-
Strong performance. MPCF achieves SOTA performance on kitti test set for single-use data.
KITTI Benchmark💪 -
More friendly. MPCF utilizes less than 7 GB during training and ~3 GB during inference. (i.e., RTX 3090, RTX4090 are enough to train our MPCF). 😀
Model Zoo
We release the model based on KITTI Dataset.
-
All models are trained with 1 RTX-3090 or 4 RTX-4090 GPUs and are available for download.
-
As for KITTI validation set, the models are trained with train split (3712 samples).
-
As for KITTI test set, please use slightly score (~0.5) threshold and train the models on all training data to achieve a desirable performance.
| Modality | GPU memory | Easy | Mod. | Hard | download | |
|---|---|---|---|---|---|---|
| mpcf-val | LiDAR+RGB | ~7GB (train)/~3 GB(val) | 95.97 | 89.67 | 86.89 | [google]() / [baidu]() |
| mpcf-test | LiDAR+RGB | ~7GB (train)/~3 GB(val) | 92.46 | 85.50 | 80.69 | [google]() / [baidu]() |
Installation
-
Prepare for the running environment.
You can follow the installation steps in
OpenPCDet. We use 1 RTX-3090 or 4 RTX-4090 GPUs to train our MPCF. -
Prepare for the data.
The dataset is follow
SFD. Anyway, you should have your dataset as follows:MPCF ├── data │ ├── kitti_pseudo │ │ │── ImageSets │ │ │── training │ │ │ ├──calib & velodyne & label_2 & image_2 & (optional: planes) & depth_dense_twise & depth_pseudo_rgbseguv_twise │ │ │── testing │ │ │ ├──calib & velodyne & image_2 & depth_dense_twise & depth_pseudo_rgbseguv_twise │ │ │── gt_database │ │ │── gt_database_pseudo_seguv │ │ │── kitti_dbinfos_train_custom_seguv.pkl │ │ │── kitti_infos_test.pkl │ │ │── kitti_infos_train.pkl │ │ │── kitti_infos_trainval.pkl │ │ │── kitti_infos_val.pkl ├── pcdet ├── tools.
-
Setup.
conda create -n MPCF_env python=3.8 conda activate MPCF_env pip install torch==2.0.0+cu117 torchvision==0.15.1+cu117 torchaudio==2.0.1+cu117 -f https://download.pytorch.org/whl/torch_stable.html (or: pip install torch==1.9.1+cu111 torchvision==0.10.1+cu111 torchaudio==0.9.1 -f https://download.pytorch.org/whl/torch_stable.html) pip install -r requirements.txt pip install spconv-cu116 (or spconv-cu111) cd MPCF python setup.py develop cd pcdet/ops/iou3d/cuda_op python setup.py develop
Getting Started
You can find the training and testing commands in tools/GP_run.sh
-
Creat kitti_pkl and GT
python -m pcdet.datasets.kitti.kitti_dataset_custom create_kitti_infos ../tools/cfgs/dataset_configs/kitti_dataset_custom.yaml -
Training. (We recommend running on single GPU, and our optimal model was trained using just 1 GPU.)
For single GPU
cd tools python train.py --gpu_id 0 --workers 1 --cfg_file cfgs/kitti_models/mpcf.yaml \ --batch_size 1 --epochs 60 --max_ckpt_save_num 25 --fix_random_seedFor 4 GPUs
cd tools python -m torch.distributed.launch --nnodes 1 --nproc_per_node=4 --master_port 25511 train.py \ --gpu_id 0,1,2,3 --launch 'pytorch' --workers 4 \ --batch_size 4 --cfg_file cfgs/kitti_models/mpcf.yaml --tcp_port 61000 \ --epochs 60 --max_ckpt_save_num 30 --fix_random_seed -
Evaluation.
cd tools python test.py --gpu_id 1 --workers 4 --cfg_file cfgs/kitti_models/mpcf_test.yaml --batch_size 1 \ --ckpt ../output/kitti_models/mpcf/default/ckpt/checkpoint_epoch_57.pth #--save_to_file
License
This code is released under the Apache 2.0 license.
Acknowledgements
We thank these great works and open-source repositories: OpenPCDet, SFD, and Voxel-RCNN.