Skip to content

Latest commit

 

History

History
118 lines (88 loc) · 5.22 KB

GETTING_START.md

File metadata and controls

118 lines (88 loc) · 5.22 KB
Raw

Getting Started with CenterPoint

Modified from det3d's original document.

Prepare data

Download data and organise as follows

# For nuScenes Dataset         
└── NUSCENES_DATASET_ROOT
       ├── samples       <-- key frames
       ├── sweeps        <-- frames without annotation
       ├── maps          <-- unused
       ├── v1.0-trainval <-- metadata

Create a symlink to the dataset root

mkdir data && cd data
ln -s DATA_ROOT 
mv DATA_ROOT nuScenes # rename to nuScenes

Remember to change the DATA_ROOT to the actual path in your system.

Create data

Data creation should be under the gpu environment.

We have created all info files and GT database which can be downloaded here and here(download both files). If you prefer to generate the files locally, please use the following script. The only difference is that our old GT database still contains annotated boxes without any radar/lidar points. However, this should not affect the training as we ensure to remove all GT objects with less than 5 lidar points during the GT-Sampling augmentation.


# nuScenes
python tools/create_data.py nuscenes_data_prep --root_path=NUSCENES_TRAINVAL_DATASET_ROOT --version="v1.0-trainval" --nsweeps=10

In the end, the data and info files should be organized as follows

# For nuScenes Dataset 
└── CenterPoint
       └── data    
              └── nuScenes 
                     ├── samples       <-- key frames
                     ├── sweeps        <-- frames without annotation
                     ├── maps          <-- unused
                     |── v1.0-trainval <-- metadata and annotations
                     |── infos_train_10sweeps_withvelo_filter_True.pkl <-- train annotations
                     |── infos_val_10sweeps_withvelo_filter_True.pkl <-- val annotations
                     |── dbinfos_train_10sweeps_withvelo.pkl <-- GT database info files
                     |── gt_database_10sweeps_withvelo <-- GT database

Train & Evaluate in Command Line

Now we only support training and evaluation with gpu. Cpu only mode is not supported.

Use the following command to start a distributed training using 4 GPUs. The models and logs will be saved to work_dirs/CONFIG_NAME

python -m torch.distributed.launch --nproc_per_node=4 ./tools/train.py CONFIG_PATH

For distributed testing with 4 gpus,

python -m torch.distributed.launch --nproc_per_node=4 ./tools/dist_test.py CONFIG_PATH --work_dir work_dirs/CONFIG_NAME --checkpoint work_dirs/CONFIG_NAME/latest.pth

For testing with one gpu and see the inference time,

python ./tools/dist_test.py CONFIG_PATH --work_dir work_dirs/CONFIG_NAME --checkpoint work_dirs/CONFIG_NAME/latest.pth --speed_test

The pretrained models and configurations are in MODEL ZOO.

Tracking

Please refer to tracking_scripts to reproduce all tracking results. The detection files are provided in the MODEL ZOO

Test Set

Organize your dataset as follows

# For nuScenes Dataset 
└── CenterPoint
       └── data    
              └── nuScenes 
                     ├── samples       <-- key frames
                     ├── sweeps        <-- frames without annotation
                     ├── maps          <-- unused
                     |── v1.0-trainval <-- metadata and annotations
                     |── infos_train_10sweeps_withvelo_filter_True.pkl <-- train annotations
                     |── infos_val_10sweeps_withvelo_filter_True.pkl <-- val annotations
                     |── dbinfos_train_10sweeps_withvelo.pkl <-- GT database info files
                     |── gt_database_10sweeps_withvelo <-- GT database 
                     └── v1.0-test <-- main test folder 
                            ├── samples       <-- key frames
                            ├── sweeps        <-- frames without annotation
                            ├── maps          <-- unused
                            |── v1.0-test <-- metadata and annotations
                            |── infos_test_10sweeps_withvelo.pkl <-- test info

Download the centerpoint_voxel_1440_dcn_flip here, save it into work_dirs/centerpoint_voxel_1440_dcn_flip_testset, then run the following commands in the main folder to get detection prediction

python tools/dist_test.py configs/centerpoint/nusc_centerpoint_voxelnet_dcn_0075voxel_flip_testset.py --work_dir work_dirs/nusc_centerpoint_voxelnet_dcn_0075voxel_flip_testset  --checkpoint work_dirs/nusc_centerpoint_voxelnet_dcn_0075voxel_flip_testset/epoch_20.pth  --speed_test

With the generated detection files, you can create the tracking prediction by running

bash tracking_scripts/centerpoint_voxel_1440_dcn_flip_testset.sh