yileccc/teri

TERI: An Effective Framework for Trajectory Recovery with Irregular Time Intervals

Dependencies

• Python>=3.7
• torch>=1.13.0

Datasets

Please download the T-drive dataset here, and then extract it in data folder.

Evaluate Model

We provide the trained models for the two stages in model folder. You can directly evaluate the trained models on test set by running:

cd recovery_stage
python test_TERI.py

Train Model

To train TERI, change to the recovery_stage or detection_stage folder for the stage you are interested and run following command:

python train.py --batch_size=128 --num_epochs=150

teri's Issues

request for dataset

The T-drive dataset can't be downloaded. Could you please send the T-drive dataset to me? Thank you very much.
My e-mail address: [email protected]