TrEP: Transformer-based Evidential Prediction for Pedestrian Intention with Uncertainty

Welcome to the official repository for the TrEP project, as presented at AAAI 2023. This project focuses on leveraging transformer-based models for predicting pedestrian intentions with an emphasis on handling uncertainty in the predictions.

Publication

The details of our approach and findings are thoroughly documented in our paper. You can access the paper through the following links:

• AAAI 2023 Proceedings: TrEP on AAAI
• ResearchGate PDF: Download PDF

Getting Started

To run the project, follow these steps:

Prerequisites

• Conda environment manager.

Setting Up the Environment

1. Clone the repository to your local machine:

   git clone https://github.com/zzmonlyyou/TrEP.git
   cd TrEP
   
   

2. Create and activate the Conda environment:

 conda env create -f environment.yml
 conda activate TrEP-env

This will install all the necessary dependencies as specified in environment.yml.

3. Training the Model

 python train.py

The arguments and their descriptions are documented within train.py. Adjust them as per your requirements.

Contact

For any queries or further discussions, please contact Zhengming Zhang.