DCENet-PyTorch

PyTorch Implementation of DCENet (https://arxiv.org/abs/2010.16267) for Trajectory Forecasting

Requirements

• python 3.8
• pytorch 1.7.1
• matplotlib
• scipy
• neptune-client if you need

Dataset

Processed data (./processed_data/train/train_merged0.npz, ./processed_data/train/train_merged1.npz, ./processed_data/train/train_merged2.npz, ./processed_data/train/biwi_hotel.npz) is requried.

You can obtain the processed data from the original repository (https://github.com/tanjatang/DCENet).

Train

• Command for training

python main.py --gpu $GPU_NUMS --config $CONFIG_FILENAME

• Example

python main.py --gpu 0 --config config.json

Test

• Command for evaluation

python evaluate.py --gpu $GPU_NUMS --config $CONFIG_FILENAME --resume-name #CHECKPOINT_FILENAME

• Example

python evaluate.py --gpu 0 --config config.json --resume-name best_model.pth

Performance on Biwi Hotel

Evaluation Results @Top25

Criteria	Original Implementation (Tensorflow)	My Implementation (PyTorch)
ADE	0.37 m	0.36 m
FDE	0.76 m	0.67 m
Hausdorff Distance	0.75 m	0.67 m
Speed Deviation	0.06 m/s	0.05 m/s
Heading Error	25.60 degree	24.67 degree

Evaluation Results for Most-likely Predictions

Criteria	Original Implementation (Tensorflow)	My Implementation (PyTorch)
ADE	0.39 m	0.42 m
FDE	0.78 m	0.79 m
Hausdorff Distance	0.77 m	0.78 m
Speed Deviation	0.06 m/s	0.05 m/s
Heading Error	30.98 degree	30.62 degree

License

Model details and most of utility functions are from from the origianl DCENet repository (https://github.com/tanjatang/DCENet).

Codes for progress bar came from https://github.com/AaronHeee/MEAL.

Codes for early stopping came from https://github.com/Bjarten/early-stopping-pytorch.

Who Am I?

I am on Ph.D course in Artificial Intelligence Lab. (Homepage), Gwangju Institute of Science and Technology (GIST, Homepage), Korea.