Abstract
The practice of yoga has been shown to have numerous benefits for both physical and mental health, and it has gained popularity worldwide as a form of exercise and relaxation. However, yoga postures can be complex and challenging, especially for beginners who may struggle with proper alignment and positioning. To address this issue, there is a need for a dataset of different yoga postures that can be used to develop computer vision algorithms capable of recognizing and analyzing yoga poses. For this we created the image and video datasets of different yoga asana using the mobile device Samsung Galaxy M30s. The dataset contains images and videos of effective (right) and ineffective postures for 10 Yoga asana, with a total of 11,344 images and 80 videos. The image dataset is organized into 10 subfolders, each with "Effective (right) Steps" and "Ineffective (wrong) Steps" folders. The video dataset has 4 videos for each posture, with 40 videos demonstrating effective (right) postures and 40 demonstrating ineffective (wrong) postures. This dataset benefits app developers, machine learning researchers, Yoga instructors, and practitioners, who can use it to develop apps, train computer vision algorithms, and improve their practice. We strongly believe that this type of dataset would provide the foundation for the development of new technologies that can help individuals improve their Yoga practice, such as posture detection and correction tools or personalized recommendations based on individual abilities and needs.