---
license: other
license_name: other
license_link: https://github.com/hukenovs/hagrid/blob/master/license/en_us.pdf
dataset_info:
  features:
  - name: image
    dtype: image
  - name: label
    dtype:
      class_label:
        names:
          '0': call
          '1': dislike
          '2': fist
          '3': four
          '4': grabbing
          '5': grip
          '6': hand_heart
          '7': hand_heart2
          '8': holy
          '9': like
          '10': little_finger
          '11': middle_finger
          '12': mute
          '13': no_gesture
          '14': ok
          '15': one
          '16': palm
          '17': peace
          '18': peace_inverted
          '19': point
          '20': rock
          '21': stop
          '22': stop_inverted
          '23': take_picture
          '24': three
          '25': three2
          '26': three3
          '27': three_gun
          '28': thumb_index
          '29': thumb_index2
          '30': timeout
          '31': two_up
          '32': two_up_inverted
          '33': xsign
  splits:
  - name: train
    num_bytes: 96145277803.94
    num_examples: 821458
  - name: validation
    num_bytes: 11767694412.8
    num_examples: 99200
  - name: test
    num_bytes: 19221400466.0
    num_examples: 165500
  download_size: 128293027578
  dataset_size: 127134372682.74
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train/train-*
  - split: validation
    path: data/val/validation-*
  - split: test
    path: data/test/test-*
---

# Dataset Description

## Dataset Summary

HaGRIDv2 is a large-scale image dataset designed for hand gesture recognition (HGR). It contains **1,086,158 FullHD RGB images** across **33 gesture classes** and an additional **"no_gesture" class**, which represents natural hand postures. This dataset is ideal for developing HGR systems for applications like video conferencing, home automation, and automotive interfaces. While composed of static images, it also supports training models for dynamic gesture recognition.

## Supported Tasks

- **Image Classification**: Classify images into one of the 33 gesture classes or the "no_gesture" class.

# Citation Information

If you use this dataset, please cite the following:

```bibtex
@misc{nuzhdin2024hagridv21mimagesstatic,
  title={HaGRIDv2: 1M Images for Static and Dynamic Hand Gesture Recognition},
  author={Anton Nuzhdin and Alexander Nagaev and Alexander Sautin and Alexander Kapitanov and Karina Kvanchiani},
  year={2024},
  eprint={2412.01508},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2412.01508},
}

@InProceedings{Kapitanov_2024_WACV,
  author = {Kapitanov, Alexander and Kvanchiani, Karina and Nagaev, Alexander and Kraynov, Roman and Makhliarchuk, Andrei},
  title = {HaGRID -- HAnd Gesture Recognition Image Dataset},
  booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
  month = {January},
  year = {2024},
  pages = {4572-4581}
}
```

# More Information

For additional details and license, visit the [GitHub repository](https://github.com/hukenovs/hagrid).