This is the official implementation of GPR , a simple yet effective ground segmentation algorithm for 3D detection tasks.
a. Clone this repository
git clone https://github.com/yhc2021/GPR && cd GPR
b. Configure the environment
We have tested this project with the following environments:
- Ubuntu18.04/20.04
- Python >= 3.7
- PyTorch = 1.10
- CUDA = 11.3
- CMake >= 3.20
*You are encouraged to try to install higher versions above, please refer to the official github repository for more information. Note that the maximum number of parallel frames during inference might be slightly decrease due to the larger initial GPU memory footprint with updated Pytorch
version.
c. Install pcdet
toolbox.
pip install -r requirements.txt
python setup.py develop
d. Prepare the datasets.
Download the official KITTI with road planes and Waymo datasets, then organize the unzipped files as follows:
GPR
├── data
│ ├── kitti
│ │ ├── ImageSets
│ │ ├── training
│ │ │ ├──calib & velodyne & label_2 & image_2 & (optional: planes)
│ │ ├── testing
│ │ ├── calib & velodyne & image_2
│ ├── waymo
│ │ │── ImageSets
│ │ │── raw_data
│ │ │ │── segment-xxxxxxxx.tfrecord
| | | |── ...
| | |── waymo_processed_data_v0_5_0
│ │ │ │── segment-xxxxxxxx/
| | | |── ...
│ │ │── waymo_processed_data_v0_5_0_gt_database_train_sampled_1/
│ │ │── waymo_processed_data_v0_5_0_waymo_dbinfos_train_sampled_1.pkl
│ │ │── waymo_processed_data_v0_5_0_gt_database_train_sampled_1_global.npy (optional)
│ │ │── waymo_processed_data_v0_5_0_infos_train.pkl (optional)
│ │ │── waymo_processed_data_v0_5_0_infos_val.pkl (optional)
├── pcdet
├── tools
Generate the data infos by running the following commands:
# KITTI dataset
python -m pcdet.datasets.kitti.kitti_dataset create_kitti_infos tools/cfgs/dataset_configs/kitti_dataset.yaml
# Waymo dataset
python -m pcdet.datasets.waymo.waymo_dataset --func create_waymo_infos \
--cfg_file tools/cfgs/dataset_configs/waymo_dataset.yaml
The configuration files are in tools/cfgs/kitti_models/GPR.yaml
and tools/cfgs/waymo_models/GPR.yaml
, and the training scripts are in tools/scripts
.
Train with single or multiple GPUs: (e.g., KITTI dataset)
python train.py --cfg_file cfgs/kitti_models/GPR.yaml
# or
sh scripts/dist_train.sh ${NUM_GPUS} --cfg_file cfgs/kitti_models/GPR.yaml
Evaluate with single or multiple GPUs: (e.g., KITTI dataset)
python test.py --cfg_file cfgs/kitti_models/GPR.yaml --batch_size ${BATCH_SIZE} --ckpt ${PTH_FILE}
# or
sh scripts/dist_test.sh ${NUM_GPUS} \
--cfg_file cfgs/kitti_models/GPR.yaml --batch_size ${BATCH_SIZE} --ckpt ${PTH_FILE}
-
This work is built upon the
OpenPCDet
(version0.5
), an open source toolbox for LiDAR-based 3D scene perception. Please refer to the official github repository for more information. -
Parts of our Code refer to IA-SSD and Patchwork++.
- Release pre-trained model of GPR on KITTI dataset
- Release pre-trained model of GPR on Waymo dataset
- Release gpu code of GPR for Waymo dataset
- Update the patchwork++ compilation documentation
This project is released under the Apache 2.0 license.