This repository contains the code and visualization demo for our paper "Diff-ExpertNet: A Deep Expert Framework Using Conditional Diffusion Models for Vessel Trajectory Generation" submitted to [KDD] 2025.
code/
: Contains the source code for this paper.data_preprocessing/
: Scripts for AIS data preprocessing.models/
: Implementation models of our proposed framework.utils/
: Utility scripts.
map/
: Map image, ports and coastlines.demo/
: Visualization demo about experimental results.
- Download public AIS dataset from source.
- Details about AIS data preprocessing can be found in
code/data_preprocessing
directory.
We merged the vessel trajectories from the test set for January 2023 into a single day, and then provided a comparison between the ground-truth and generated results.