Awesome

[AAAI'2024] - EVPTrack

The official implementation for the AAAI 2024 paper [Explicit Visual Prompts for Visual Object Tracking].

[Models], [Raw Results], [Training logs]

Highlights

:star2: New Explicit Visual Prompts-base Tracking Framework

EVPTrack is a simple and high performance tracker. It achieves SOTA performance on multiple benchmarks by utilizing spatio-temporal and multi-scale template information.

:star2: Strong Performance

Install the environment

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

Data Preparation

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

${PROJECT_ROOT}
 -- data
     -- lasot
         |-- airplane
         |-- basketball
         |-- bear
         ...
     -- got10k
         |-- test
         |-- train
         |-- val
     -- coco
         |-- annotations
         |-- images
     -- trackingnet
         |-- TRAIN_0
         |-- TRAIN_1
         ...
         |-- TRAIN_11
         |-- TEST

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

Training

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

python tracking/train.py \
--script evptrack --config EVPTrack-full-224 \
--save_dir ./output \
--mode multiple --nproc_per_node 4 \
--use_wandb 0

Replace --config with the desired model config under experiments/evptrack.

We use wandb to record detailed training logs, in case you don't want to use wandb, set --use_wandb 0.

Test and Evaluation

• LaSOT or other off-line evaluated benchmarks (modify --dataset correspondingly)

python tracking/test.py --tracker_param EVPTrack-full-224 --dataset lasot --threads 8 --num_gpus 4
python tracking/analysis_results.py # need to modify tracker configs and names

• GOT10K-test

python tracking/test.py  --tracker_param EVPTrack-full-224 --dataset got10k --threads 8 --num_gpus 4

• TrackingNet

python tracking/test.py  --tracker_param EVPTrack-full-224 --dataset trackingnet --threads 8 --num_gpus 4

Test FLOPs, and Speed

Note: The speeds reported in our paper were tested on a single RTX2080Ti GPU.

python tracking/profile_model.py --script evptrack --config baseline

Acknowledgments

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

Citation

If our work is useful for your research, please consider citing:

@inproceedings{shi2024evptrack,
  title={Explicit Visual Prompts for Visual Object Tracking}, 
  author={Liangtao Shi and Bineng Zhong and Qihua Liang and Ning Li and Shengping Zhang and Xianxian Li},
  booktitle={AAAI},
  year={2024}
}