Bhavani-23/Ocularone-Hazard-Vest-Dataset

Ocularone: Hazard Vest Dataset

🤗 Hugging Face   |   🚀 Github   

Note: The dataset was used to train YOLO models (v8 and v11) to detect the hazard vest on a person. The models can be found here.

The dataset has two files: one which contains all the images that were collected by a drone of a person wearing a hazard vest and walking around our University campus; the other file contains the training data with images that were randomly picked from the total dataset.

Dataset Structure

The dataset consists of multiple subfolders which describe the categories/conditions under which the images were collected. Each category folder has multiple subfolders for the images.

References

@misc{raj2025ocularonebenchbenchmarkingdnnmodels,
      title={Ocularone-Bench: Benchmarking DNN Models on GPUs to Assist the Visually Impaired}, 
      author={Suman Raj and Bhavani A Madhabhavi and Kautuk Astu and Arnav A Rajesh and Pratham M and Yogesh Simmhan},
      year={2025},
      eprint={2504.03709},
      archivePrefix={arXiv},
      primaryClass={cs.DC},
      url={https://arxiv.org/abs/2504.03709}, 
}