this repository provides Code for ijcai2023 paper: Open Anomalous Trajectory Recognition via Probabilistic Metric Learning
The datasets used in our paper are collected by Kaggle and DiDi Chuxing. Please download from http://outreach.didichuxing.com/research/opendata and https://www.kaggle.com/c/pkdd-15-predict-taxi-service-trajectory-i/data respectively.
Run data_process.py to generate and load anomalous trajectories. Run model.py to train and test our model.
- python = 3.7
- pytorch = 1.8
- numpy
- scikit-learn