Skip to content

jcwang0602/UITrack

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

UITrack

Install the environment

Use the Anaconda

conda create -n uitrack python=3.8
conda activate uitrack
bash install.sh

Data Preparation

Put the tracking datasets in ./data. It should look like:

${UITrack_ROOT}
 -- data
     -- LaSOTBenchmark
         |-- airplane
         |-- basketball
         |-- bear
         ...
     -- tnl2k
         -- train
            |-- Arrow_Video_ZZ04_done
            |-- Assassin_video_1-Done
            ...
         -- test
            |-- advSamp_Baseball_game_002-Done
            |-- advSamp_Baseball_video_01-Done
            ...
     -- OTB_sentences
         |-- OTB_query_test
         |-- OTB_query_train
         |-- OTB_videos
     -- refcoco
         -- annotations
            -- refcoco-unc
               |-- instances.json
               |-- ix_to_token.pkl
               ...
            -- refcocog-google
            ...
         |-- images
             |--train2014
             |--train2017

Set project paths

Run the following command to set paths for this project

python tracking/create_default_local_file.py --workspace_dir . --data_dir ./data --save_dir output

After running this command, you can also modify paths by editing these two files

lib/train/admin/local.py  # paths about training
lib/test/evaluation/local.py  # paths about testing

Train UITrack

Download pre-trained MAE ViT-Base weights and put it under $PROJECT_ROOT$/pretrained_models (different pretrained models can also be used, see MAE for more details).

Training with multiple GPUs using DDP.

bash train.sh

Test and evaluate UITrack on benchmarks

Download the model weights and raw results from Baidu Netdisk.

  • LaSOT/TNL2K/OTB99-L. More details of test settings can be found at
bash test.sh

Compute FLOPs/Params and test speed

python tracking/profile_model.py --config="uitrack_256_mae_32x4_ep100_prompt"
python tracking/profile_model.py --config="uitrack_384_mae_32x4_ep100_prompt"

Acknowledgments

  • Thanks for the OSTrack and MAE library, which helps us to quickly implement our ideas.

About

[ICME 2025] Overcoming Feature Contamination by Unidirectional Information Modeling for Vision-Language Tracking

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages