Tutorial for SLT-TransT & SLT-TrDiMP (Based on PyTracking)

SLT-TransT and SLT-TrDiMP are implemented based on PyTracking library, which is composed of LTR for training and pytracking for testing.

Training

For the detailed usage of the LTR library, please refer to LTR.

Modify ltr/admin/local.py to set the paths to datasets, results, etc.
Download the baseline model (e.g., transt.pth) from [Models] and save it in your local path.
(Optional) You can also train the baseline model by yourself by running the following command:
```
python ltr/run_training.py transt transt
```
Run the following commands to train the SLT tracker. Note that you should specify the path to the pretrained model in your training setting, e.g., slt_transt.py.
```
python ltr/run_training.py slt_transt slt_transt
```

For the detailed usage of the pytracking library, please refer to pytracking.

To test the tracker, run the following command:

python pytracking/run_tracker.py [tracker_name] [parameter_name] --dataset_name [dataset_name]

For example,

python pytracking/run_tracker.py slt_transt slt_transt --dataset_name lasot

Tip: run with --threads [num_threads] and --num_gpu [num_gpus] for multi-gpu multi-threads inference.

To see the evaluation results, run the following command:

python pytracking/show_results.py [tracker_name] [parameter_name] --dataset_name [dataset_name]

For example,

python pytracking/show_results.py slt_transt slt_transt --dataset_name lasot

To submit the results on evaluation servers (e.g., TrackingNet and GOT-10k), use the scripts in util_scripts.