This repo contains the official implementation of our paper: "End-to-End Trajectory Distribution Prediction Based on Occupancy Grid Maps". Ke Guo, Wenxi Liu, Jia Pan.

CVPR 2022
paper

Installation

Environment

• Python >= 3.7
• PyTorch == 1.8.0

Data and pretrained model

Please download the pretrained model and data from onedrive(https://connecthkuhk-my.sharepoint.com/:u:/g/personal/u3006612_connect_hku_hk/EXqC6hjGTphKh8TkjrwtByEB3FFZ_dpCu0Rs6N7CTG2gag?e=5q4Knz). Extract the zip file into the main folder.

Data Preprocessing

Here is the detail of data preprocessing. You can skip it by using the data from google drive.

• SDD (Trajnet split)

1. Download the Trajnet split data from Y-Net. Put the data under data/SDD

2. Run script to process the downloaded "train_trajnet.pkl" and "test_trajnet.pkl":

   python data/SDD/process_trajnet.py
   

• SDD(P2T split)

1. Download the P2T split data from P2T. Put the data under data/SDD

2. Run script to process the downloaded "SDDtrain.mat", "SDDval.mat" and "SDDtest.mat":

   python data/SDD/process_p2t.py
   

• inD

1. Obtain the processed inD data from Y-Net. Put the data under data/SDD

2. Run script to process the downloaded "inD_train.pickle" and "inD_test.pickle":

   python data/SDD/process_inD.py
   

Training

Training the model for Trajnet:

  ```
  python train.py  --dataset "trajnet"
  ``` 

For SDD(p2t split) or inD, the "trajnet" need to be replaced by "sdd" or "ind".

Evaluation

Evaluating on Trajnet dataset:

  ```
  python eval.py  --dataset "trajnet"
  ``` 

For SDD(p2t split) or inD, the "trajnet" need to be replaced by "sdd" or "ind".

Citation

@inproceedings{guo2022end,
  title={End-to-End Trajectory Distribution Prediction Based on Occupancy Grid Maps},
  author={Ke, Guo and Wenxi, Liu and Jia, Pan},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  year={2022}
}