Skip to content

code and materials for paper "Multi-Agent Trajectory Predition Based on Graph Neural Network", IEEE RCAR 2021

Notifications You must be signed in to change notification settings

LeoDuhz/MultiAgentTrajPrediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 

Repository files navigation

RCAR2021_MultiAgentTrajPrediction

code and materials for Multi-Agent Trajectory Predition Based on Graph Neural Network

Note:This repo stops to be updated, and for further work of us, you can switch to this repo which keeps on updating.

Heterogeneous GNN

  • main.py: main function of LOG Analysis model training process

  • parameters.py: parameters management of the training process

  • dataFormat.py: basic data format of player, ball and game data

  • referee.py: referee command defined in proto file

  • dataPreprocess.py: preprocess the text file we get from our vision module, converting into formatted data we can use and doing Min-Max normalization

  • SSLDataset.py: construct the graph structure for future gnn training

  • mys2v.py: basic graph neural network we use

  • pna/mypna.py: Heterogeneous PNAConv

  • heterogeneous/myheter.py: Heterogeneous graph neural network frame supporting different GNNs

  • Net.py: neural network we construct

  • debug/debug.py: draw gradient of graph neural network

  • visualize.py: draw pictures of our training result

  • testOurModel.py: load torch model and see results and also visualiztion

Homogeneous GNN

About

code and materials for paper "Multi-Agent Trajectory Predition Based on Graph Neural Network", IEEE RCAR 2021

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages