Awesome

Updates

[22.12.2024] Check out our follow-up work SPAM [ECCV 2024]! We are releasing an upgraded version of SUSHI code base, fully integrated into the SPAM code base in our new repository.

Unifying Short and Long-Term Tracking with Graph Hierarchies :sushi:

Official repository of our CVPR 2023 paper

Unifying Short and Long-Term Tracking with Graph Hierarchies

Orcun Cetintas*, Guillem Brasó*, Laura Leal-Taixe

Abstract

Tracking objects over long videos effectively means solving a spectrum of problems, from short-term association for un-occluded objects to long-term association for objects that are occluded and then reappear in the scene. Methods tackling these two tasks are often disjoint and crafted for specific scenarios, and top-performing approaches are often a mix of techniques, which yields engineering-heavy solutions that lack generality. In this work, we question the need for hybrid approaches and introduce SUSHI, a unified and scalable multi-object tracker. Our approach processes long clips by splitting them into a hierarchy of subclips, which enables high scalability. We leverage graph neural networks to process all levels of the hierarchy, which makes our model unified across temporal scales and highly general. As a result, we obtain significant improvements over state-of-the-art on four diverse datasets.

Results

MOTChallenge - Test Set

DanceTrack - Test Set

BDD - Test Set

Setup

1. Clone and enter this repository

   git clone https://github.com/dvl-tum/SUSHI.git
   cd SUSHI
   

2. Create an Anaconda environment for this project:

   conda env create -f environment.yml
   conda activate SUSHI
   

3. Clone fast-reid (latest version should be compatible but we use this version) and install its dependencies. The fast-reid repo should be inside SUSHI root:

   SUSHI
   ├── src
   ├── fast-reid
   └── ...
   

4. Download re-identification model we use from fast-reid and move it inside SUSHI/fastreid-models/model_weights/

5. Download MOT17, MOT20 and DanceTrack datasets. In addition, prepare seqmaps to run evaluation (for details see TrackEval). We provide an example seqmap. Overall, the expected folder structure is:

   DATA_PATH
   ├── DanceTrack
   │   └── ...
   ├── MOT20
   │   └── ...
   └── MOT17
       └── seqmaps
       │    ├── seqmap_file_name_matching_split_name.txt
       │    └── ...
       └── train
       │    ├── MOT17-02
       │    │   ├── det
       │    │   │   └── det.txt
       │    │   └── gt
       │    │   │   └── gt.txt
       │    │   └── img1 
       │    │   │   └── ...
       │    │   └── seqinfo.ini
       │    └── ...
       └── test
            └── ...
   
   

6. (OPTIONAL) Download our detections and pretrained models and arrange them according to the expected file structure (See step 5).

Training

You can launch a training from the command line. An example command for MOT17 training:

RUN=example_mot17_training
REID_ARCH='fastreid_msmt_BOT_R50_ibn'

DATA_PATH=YOUR_DATA_PATH

python scripts/main.py --experiment_mode train --cuda --train_splits mot17-train-all --val_splits mot17-train-all --run_id ${RUN}_${REID_ARCH} --interpolate_motion --linear_center_only --det_file byte065 --data_path ${DATA_PATH} --reid_embeddings_dir reid_${REID_ARCH} --node_embeddings_dir node_${REID_ARCH} --zero_nodes --reid_arch $REID_ARCH --edge_level_embed --save_cp

You can find specific commands and configurations here.

Testing

You can test a trained model from the command line. An example for testing on MOT17 training set:

RUN=example_mot17_test
REID_ARCH='fastreid_msmt_BOT_R50_ibn'

DATA_PATH=your_data_path
PRETRAINED_MODEL_PATH=your_pretrained_model_path

python scripts/main.py --experiment_mode test --cuda --test_splits mot17-train-all --run_id ${RUN}_${REID_ARCH} --interpolate_motion --linear_center_only --det_file byte065 --data_path ${DATA_PATH} --reid_embeddings_dir reid_${REID_ARCH} --node_embeddings_dir node_${REID_ARCH} --zero_nodes --reid_arch $REID_ARCH --edge_level_embed --save_cp --hicl_model_path ${PRETRAINED_MODEL_PATH}

Launching this experiment with our MOT17 Private model should result in: 90.826 IDF1, 80.884 HOTA and 91.953 MOTA.

You can find specific commands and configurations here.

Citation

If you use our work in your research, please cite our publication:

@InProceedings{Cetintas_2023_CVPR,
    author    = {Cetintas, Orcun and Bras\'o, Guillem and Leal-Taix\'e, Laura},
    title     = {Unifying Short and Long-Term Tracking With Graph Hierarchies},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2023},
    pages     = {22877-22887}
}

Acknowledgements

We use the codebase of TrackEval for evaluation. We thank the authors for their great work!