hawk icon indicating copy to clipboard operation
hawk copied to clipboard

Hawk: Learning to Understand Open-World Video Anomalies

[NeurIPS 2024] Hawk: Learning to Understand Open-World Video Anomalies

This is the official repository for Hawk.

Jiaqi Tang^, Hao Lu^, Ruizheng Wu, Xiaogang Xu, Ke Ma, Cheng Fang,

Bin Guo, Jiangbo Lu, Qifeng Chen and Ying-Cong Chen*

^: Equal contribution. *: Corresponding Author.

GitHub license made-for-VSCode Visits Badge

Have eyes like a HAWK!

πŸ” Motivation - Have eyes like a Hawk!

  • 🚩 Current VAD systems are often limited by their superficial semantic understanding of scenes and minimal user interaction.

  • 🚩 Additionally, the prevalent data scarcity in existing datasets restricts their applicability in open-world scenarios.

    Hawk

πŸ“’ Updates

  • βœ… Step 26, 2024 - Hawk is accepted by NeurIPS 2024.
  • βœ… July 29, 2024 - We release the dataset of Hawk. Check this Google Cloud link for DOWNLOAD.

▢️ Getting Started

πŸͺ’ Installation

πŸ’Ύ Dataset Preparation

  • DOWNLOAD all video datasets for their original sources.

    1. CUHK_Avenue
    2. DoTA
    3. Ped1
    4. Ped2
    5. ShanghaiTech
    6. UBNormal
    7. UCF_Crime
  • Google Drive Link to DOWNLOAD our annotations.

  • Data Structure: each folder contains one annotation file (e.g. CUHK Avenue, DoTA, etc.). The All_Mix directory contains all of the datasets in training and testing.

  • The dataset is organized as follows:

    data
    β”œβ”€β”€ All_Mix
    β”‚   β”œβ”€β”€ all_videos_all.json
    β”‚   β”œβ”€β”€ all_videos_test.json
    β”‚   └── all_videos_train.json
    β”‚    
    β”œβ”€β”€ CUHK_Avenue
    β”‚   └── Avenue.json
    β”œβ”€β”€ DoTA
    β”‚   └── DoTA.json
    β”œβ”€β”€ Ped1
    β”‚   β”œβ”€β”€ ...
    β”œβ”€β”€ ...
    └── UCF_Crime
        └── ...
    

    Note:the data path should be redefined.

🏰 Pretrained Model

πŸ”¨ Configuration

⏳ Testing

πŸ–₯️ Training

⚑ Performance

🌐 Citations

The following is a BibTeX reference:

@inproceedings{tang2024hawk,
      title={Hawk: Learning to Understand Open-World Video Anomalies}, 
      author={Jiaqi Tang and Hao Lu and Ruizheng Wu and Xiaogang Xu and Ke Ma and Cheng Fang and Bin Guo and Jiangbo Lu and Qifeng Chen and Ying-Cong Chen},
      booktitle={Hawk: Learning to Understand Open-World Video Anomalies},
      year={2024}
}

πŸ“§ Connecting with Us?

If you have any questions, please feel free to send email to [email protected].