EtherStone/bharatanatyam-mudra-dataset

---
license: mit
task_categories:
- image-classification
- zero-shot-image-classification
language:
- en
tags:
- bharatanatyam
- mudra
- hand-gestures
- indian-classical-dance
- computer-vision
size_categories:
- 10K<n<100K
---

# Bharatanatyam Mudra Dataset

## Dataset Description

The Bharatanatyam Mudra Dataset contains **28,431 images** of hand gestures (mudras) from Bharatanatyam, a classical Indian dance form. The dataset was collected from 15 volunteers in a studio environment and includes both single-hand and double-hand gestures.

### Dataset Statistics
- **Total Images**: 28,431
- **Single Hand Gestures (Asamyukta Hastas)**: 15,396 images across 29 classes
- **Double Hand Gestures (Samyukta Hastas)**: 13,035 images across 21 classes
- **Total Classes**: 50 different mudras

## Dataset Structure

The dataset is organized into 50 classes representing different mudras:

### Single Hand Gestures (Asamyukta Hastas) - 29 classes
- Pathaka, Tripathaka, Ardhapathaka, Mayura, Katrimukha
- Ardhachandran, Aralam, Shukatundam, Mushti, Sikharam
- Kapith, Katakamukha_1, Katakamukha_2, Katakamukha_3, Suchi
- Chandrakala, Padmakosha, Sarpasirsha, Mrigasirsha, Simhamukham
- Kangulam, Alapadmam, Mukulam, Chaturam, Bramaram
- Hamsasyam, Hamsapaksham, Tamarachudam, Trishulam

### Double Hand Gestures (Samyukta Hastas) - 21 classes
- Anjali, Kapotham, Karkatta, Swastikam, Pushpaputam
- Shivalinga, Katakavardhana, Kartariswastika, Sakata, Shanka
- Chakra, Samputa, Pasha, Kilaka, Matsya
- Kurma, Varaha, Garuda, Nagabandha, Khatva, Berunda

## Data Fields

- `image`: PIL Image of the mudra
- `label`: String label of the mudra name
- `label_id`: Numerical ID for the label
- `gesture_type`: Either "single_hand" or "double_hand"

## Usage

```python
from datasets import load_dataset

# Load the dataset
dataset = load_dataset("samarth/bharatanatyam-mudra-dataset")

# Access the data
train_data = dataset["train"]
print(f"Number of samples: {len(train_data)}")
print(f"Features: {train_data.features}")

# Example: Get first image and label
sample = train_data[0]
image = sample["image"]
label = sample["label"]
print(f"Label: {label}")
```

## Applications

This dataset can be used for:
- Hand gesture recognition and classification
- Cultural heritage preservation through AI
- Computer vision research on hand pose estimation
- Educational applications for learning Bharatanatyam
- Transfer learning for other hand gesture datasets

## Citation and Acknowledgments

This data was collected as part of Ph.D. work done under the guidance of **Dr. Sunil T.T**, Professor, College of Engineering, Attingal, Thiruvananthapuram, Kerala, India.

For original-sized images or additional information, please contact: **Jisha Raj R** at [e-mail]

## License

This dataset is available under the MIT License.

## Ethical Considerations

This dataset was collected with the consent of volunteers in a controlled studio environment. The dataset represents traditional Indian cultural practices and should be used respectfully, particularly in research and educational contexts.