DeFlow: Decoder of Scene Flow Network in Autonomous Driving

Will present in ICRA'24.

Task: Scene Flow Estimation in Autonomous Driving. Pre-trained weights for models are available in Onedrive link. Check usage in 2. Evaluation or 3. Visualization.

Scripts quick view in our scripts:

• dataprocess/extract_*.py : pre-process data before training to speed up the whole training time. [Dataset we included now: Argoverse 2, more on the way: Waymo and Nuscenes, custom data.]

• 1_train.py: Train the model and get model checkpoints. Pls remember to check the config.

• 2_eval.py : Evaluate the model on the validation/test set. And also upload to online leaderboard.

• 3_vis.py : For visualization of the results with a video.

0. Setup

Environment: Clone the repo and build the environment, check detail installation for more information. Conda/Mamba is recommended.

git clone --recursive https://github.com/KTH-RPL/DeFlow.git
cd DeFlow
mamba env create -f environment.yaml

mmcv:

mamba activate deflow
cd ~/DeFlow/mmcv && export MMCV_WITH_OPS=1 && export FORCE_CUDA=1 && pip install -e .

Or another environment setup choice is Docker which isolated environment, you can pull it by. If you have different arch, please build it by yourself cd DeFlow && docker build -t zhangkin/deflow by going through build-docker-image section.

# option 1: pull from docker hub
docker pull zhangkin/deflow

# run container
docker run -it --gpus all -v /dev/shm:/dev/shm -v /home/kin/data:/home/kin/data --name deflow zhangkin/deflow /bin/zsh

1. Train

Download tips in dataprocess/README.md

Prepare Data

Normally need 10-45 mins finished run following commands totally (my computer 15 mins, our cluster 40 mins).

python dataprocess/extract_av2.py --av2_type sensor --data_mode train --argo_dir /home/kin/data/av2 --output_dir /home/kin/data/av2/preprocess
python dataprocess/extract_av2.py --av2_type sensor --data_mode val --mask_dir /home/kin/data/av2/3d_scene_flow
python dataprocess/extract_av2.py --av2_type sensor --data_mode test --mask_dir /home/kin/data/av2/3d_scene_flow

Train Model

All local benchmarking methods and ablation studies can be done through command with different config, check assets/slurm for all the commands we used in our experiments.

Best fine-tuned model train with following command by other default config in conf/config.yaml and conf/model/deflow.yaml, if you will set wandb_mode=online, maybe change all entity="kth-rpl" to your own account name.

python 1_train.py model=deflow lr=2e-6 epochs=50 batch_size=16

Benchmarking and baseline methods:

python 1_train.py model=fastflow3d lr=2e-6 epochs=50 batch_size=16
python 1_train.py model=deflow lr=2e-6 epochs=50 batch_size=16

# for nsfp no need train but optimize iteration running
python 2_eval.py model=nsfp 
python 2_eval.py model=fast_nsfp

To help community benchmarking, we provide our weights including fastflow3d, deflow Onedrive link. These checkpoints also include parameters and status of that epoch inside it. If you are interested in weights of ablation studies, please contact us.

2. Evaluation

You can view Wandb dashboard for the training and evaluation results or run/submit to av2 leaderboard to get official results.

Since in training, we save all hyper-parameters and model checkpoints, the only thing you need to do is to specify the checkpoint path. Remember to set the data path correctly also.

python 2_eval.py checkpoint=/home/kin/model.ckpt av2_mode=val # it will directly prints all metric
python 2_eval.py checkpoint=/home/kin/model.ckpt av2_mode=test # it will output the av2_submit.zip for you to submit to leaderboard

Check all detailed result files (presented in our paper Table 1) in this discussion.

To submit to the Online Leaderboard, the last step will tell you the resulting path, copy it here:

# you will find there is a av2_submit.zip in the folder now. since the env is different and conflict we set new one:
mamba create -n py37 python=3.7
mamba activate py37
pip install "evalai"

# Step 2: login in eval and register your team
evalai set_token <your token>

# Step 3: Submit to leaderboard
evalai challenge 2010 phase 4018 submit --file av2_submit.zip --large --private

3. Visualization

We provide a script to visualize the results of the model. You can specify the checkpoint path and the data path to visualize the results. The step is quickly similar to evaluation.

python 3_vis.py checkpoint=/home/kin/model.ckpt dataset_path=/home/kin/data/av2/preprocess/sensor/vis

# Then terminal will tell you the command you need run. For example here is the output of the above:
Model: DeFlow, Checkpoint from: /home/kin/model_zoo/deflow.ckpt
We already write the estimate flow: deflow into the dataset, please run following commend to visualize the flow. Copy and paste it to your terminal:
python tests/scene_flow.py --flow_mode 'deflow' --data_dir /home/kin/data/av2/preprocess/sensor/mini
Enjoy! ^v^ ------ 

# Then run the command in the terminal:
python tests/scene_flow.py --flow_mode 'deflow' --data_dir /home/kin/data/av2/preprocess/sensor/mini

Note: ego_motion already compensated, so the visualization is more clear.

DeFlow_Demo.mp4

Cite & Acknowledgements

@article{zhang2024deflow,
  author={Zhang, Qingwen and Yang, Yi and Fang, Heng and Geng, Ruoyu and Jensfelt, Patric},
  title={DeFlow: Decoder of Scene Flow Network in Autonomous Driving},
  journal={arXiv preprint arXiv:2401.16122},
  year={2024}
}

This implementation is based on codes from several repositories. Thanks to these authors who kindly open-sourcing their work to the community. Please see our paper reference part to get more information. Thanks to Kyle Vedder (ZeroFlow) who kindly discussed their results with us and HKUST Ramlab's member: Jin Wu who gave constructive comments on this work. The computations were enabled by the supercomputing resource Berzelius provided by National Supercomputer Centre at Linköping University and the Knut and Alice Wallenberg Foundation, Sweden.

❤️: ZeroFlow, NSFP, FastNSF. Others good code style and tools: forecast-mae, kiss-icp