neilrigaud/hagrid-subset

HaGRID Gesture Recognition Subset

Dataset Description

A curated subset of the HaGRID (Hand Gesture Recognition Image Dataset) containing 24 gesture classes for training gesture recognition models.

Dataset Summary

• Total Images: 19,200
• Gesture Classes: 24
• Samples per Class: 800
• Image Format: JPEG
• Average Image Size: ~302 KB

Splits

Split	Images	Percentage
Train	14,592	76%
Val	1,728	9%
Test	2,880	15%

Gesture Classes

1. call
2. dislike
3. fist
4. four
5. grabbing
6. grip
7. like
8. middle_finger
9. mute
10. no_gesture
11. ok
12. one
13. palm
14. peace
15. peace_inverted
16. point
17. rock
18. stop
19. stop_inverted
20. three
21. three2
22. three3
23. two_up
24. two_up_inverted

Dataset Structure

data/
├── train/
│   ├── call/          (608 images)
│   ├── dislike/       (608 images)
│   └── ...
├── val/
│   └── ...
└── test/
    └── ...
annotations.csv         (19,200 entries)

Usage

Loading with Hugging Face Datasets

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("YOUR_USERNAME/hagrid-subset")

# Access splits
train_data = dataset["train"]
val_data = dataset["validation"]
test_data = dataset["test"]

Loading with Pandas

import pandas as pd
from PIL import Image

df = pd.read_csv("hf://datasets/YOUR_USERNAME/hagrid-subset/annotations.csv")
train_df = df[df['split'] == 'train']

Citation

If you use this dataset, please cite the original HaGRID dataset:

@inproceedings{hagrid2022,
  title={HaGRID--HAnd Gesture Recognition Image Dataset},
  author={Kapitanov, Alexander and Kvanchiani, Karina and Nagaev, Alexander and Kraynov, Andrey and Makhliarchuk, Maxim},
  booktitle={2022 International Conference on Robotics and Artificial Intelligence (ICRAI)},
  year={2022}
}

Dataset Creation

This subset was created using stratified sampling to maintain train/val/test ratios while ensuring class balance.

Validation Status: ✅ Fully validated with 100% annotation coverage

License

MIT License (inherited from original HaGRID dataset)