DenseTNT

Paper | Webpage

• This is the official implementation of the paper: DenseTNT: End-to-end Trajectory Prediction from Dense Goal Sets (ICCV 2021).
• DenseTNT v1.0 was released in November 1st, 2021.

Quick Start

Requires:

• Python 3.6+
• pytorch 1.6+

1) Install packages

 pip install -r requirements.txt

2) Install Argoverse API

https://github.com/argoai/argoverse-api

3) Compile Cython

Compile a .pyx file into a C file using Cython:

⚠️Recompiling is needed every time the pyx files are changed.

cd src/
cython -a utils_cython.pyx && python setup.py build_ext --inplace

Performance

Results on Argoverse motion forecasting validation set:

	minADE	minFDE	Miss Rate
DenseTNT w/ 100ms optimization	0.80	1.27	7.0%
DenseTNT w/ 100ms optimization (minFDE)	0.73	1.05	9.8%
DenseTNT w/ goal set predictor (online)	0.82	1.37	7.0%

Models

Path to models

DenseTNT

Train

# for carla
bash train_carla.sh
# for argoverse
bash train.sh

Evaluate

bash val.sh
# for optimize mFDE
bash val_MRmFDE.sh

Comparision

bash comparision.sh

Inference

bash inference.sh

Citation

If you find our work useful for your research, please consider citing the paper:

@inproceedings{densetnt,
  title={Densetnt: End-to-end trajectory prediction from dense goal sets},
  author={Gu, Junru and Sun, Chen and Zhao, Hang},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={15303--15312},
  year={2021}
}