Pranayama Dataset: A Collection of Breathing Exercise Visuals (Images and Videos) for Health and Wellness

Published: 14 May 2023| Version 2 | DOI: 10.17632/p2dhvhcw27.2
, Zainab Parekh,


The "Pranayama Dataset" is a diverse collection of visuals that features five different pranayama postures, including Anulom Vilom, Brahmari, Chandra Bhedan, Surya Bhedan, and Shitali Pranayama. The dataset contains both correct and incorrect postures images and videos, providing a comprehensive resource for practitioners and researchers alike. What sets this dataset apart is its diversity in terms of environmental and lighting conditions, as well as the various angles from which the images and videos were captured. This allows users to train machine learning models that can better identify and differentiate between correct and incorrect postures in a wide range of settings. With a total of 3150 images and 80 videos, this dataset is an invaluable resource for machine learning users looking to develop algorithms that can accurately recognize pranayama postures. The dimensions 256 × 256 of the images provide clear visual demonstrations, making it easy for users to train their models effectively. Overall, the "Pranayama Dataset" is an essential resource for anyone interested in using machine learning to improve their pranayama practice or for researchers looking to better understand the benefits of breathing exercises for health and wellness.



Mental Health, Sport, Yoga, Health, Posture, Physical Exercise