Advancements in Yoga Pose Recognition and Correction: A Comprehensive Literature Review
B K Pavankumar, G. Mahitha · International Journal of Soft Computing and Engineering · 2025
The growing popularity of yoga, especially in post-pandemic wellness trends, has led to an increasing demand for automated systems capable of real-time pose detection, classification, and correction. This literature review surveys and compares recent advancements in yoga pose recognition and correction technologies, with an emphasis on the integration of deep learning, computer vision, and optimisation techniques. The reviewed works employ a variety of methods, ranging from CNNs, GRUs, and Vision Transformers to heuristic-based models and hybrid architectures, for both static and dynamic pose estimation. Systems leveraging lightweight models, such as MoveNet, as well as multimodal approaches combining AR, personalised recommendations, and real-time corrective feedback, demonstrate significant potential for mobile, web-based, and wearable deployments. This review synthesizes insights on model performance, technological innovations, and future opportunities, providing a foundation for researchers and developers aiming to build intelligent, user-centric yoga tutor applications.