A Comprehensive Review of Real-Time Yoga Posture Correction Using Pose Detection and CNN–LSTM Models

Author

Mohd Salauddin, Sanjay Singh, Sneha, Dr. Pardeep Chouksey, Praveen Sadotra, Mayank Chopra

Abstract

 AI and ML have revolutionized RTYPD and corrective feedback systems by enabling fast evolution in real-time.AI and ML have helped to revolutionize RTYPD and corrective feedback systems, enabling rapid evolution in real time.The focus of this review is on the studies published since 2019 up to 2025. When dis-cussing recognizing yoga poses, artificial intelligence plays a significant role, and making real-time inferences, and even fixing mistakes are at the heart of this technique. This document critically reviews 15 peer-reviewed studies on the topic of yoga pose recognition and surveys the variety of approaches that have been under-taken so far such as the Keypoint-based estimation of poses, OpenPose, MediaPipe, PoseNet and MoveNet, deep-learning models using CNNs and CNN-LSTM hybrids, 3D-CNNs and two-stream architectures, and the angle-based correction frameworks. The findings indicate that deep learning models that are coupled with lightweight pose detectors can achieve high accuracy and real-time execution of mobile and edge devices. However, there are still many areas that need to be improved, such as the range of datasets being signifi-cantly imbalanced, generalization, inter-body robustness and corrective feedback reliability. Moreover, the majority of the current systems are based on the identification of postures and not the actual biomechanical correction. This is a review of the current methods and a comparison of performance trends. It also find out research gaps and highlights future directions. They range from multimodal sensing to feedback systems with AR/VR, datasets with standardized benchmark, and explainable correction models. The findings are provided to facilitate additional research into intelligent systems for assisting with yoga and to help advance AI-based wellness technologies.

Keywords

yoga posture detection, real-time pose estimation, machine learning, deep learning

Full Text:

Download Paper PDF

References

[1] F. B. Ashraf, M. U. Islam, M. R. Kabir, and J. Uddin, “YoNet: A Neural Network for Yoga Pose Classification,” SN Computer Science, vol. 4, no. 2, p. 198, Feb. 2023.

[2] R. Bansal, R. Sharma, P. Jain, R. Arora, S. Pal, and . Vishal, “DeepYoga: Enhancing Practice with a Real-Time Yoga Pose Recognition System,” Engineering, Technology & Applied Science Research, vol. 14, no. 6, pp. 17 704–17 710, Dec. 2024.

[3] S.-C. Yeh and C.-K. Yang, “Yoga pose recognition and motion analysis for a home-based fitness mon-itoring and health management system,” Signal, Image and Video Processing, vol. 19, no. 10, p. 841, Oct. 2025.

[4] A. Upadhyay, N. K. Basha, and B. Ananthakrishnan, “Deep Learning-Based Yoga Posture Recognition Using the Y PN-MSSD Model for Yoga Practitioners,” Healthcare, vol. 11, no. 4, p. 609, Feb. 2023.

[5] S. K. Yadav, A. Agarwal, A. Kumar, K. Tiwari, H. M. Pandey, and S. A. Akbar, “YogNet: A two-stream network for realtime multiperson yoga action recognition and posture correction,” Knowledge-Based Systems, vol. 250, p. 109097, Aug. 2022.

[6] S. Jain, A. Rustagi, S. Saurav, R. Saini, and S. Singh, “Three-dimensional CNN-inspired deep learning architecture for Yoga pose recognition in the real-world environment,” Neural Computing and Applica-tions, vol. 33, no. 12, pp. 6427–6441, Jun. 2021.

[7] S. K. Yadav, A. Agarwal, A. Kumar, K. Tiwari, H. M. Pandey, and S. A. Akbar, “YogNet: A two-stream network for realtime multiperson yoga action recognition and posture correction,” Knowledge-Based Systems, vol. 250, p. 109097, Aug. 2022.

[8] S. K. Yadav, A. Singh, A. Gupta, and J. L. Raheja, “Real-time Yoga recognition using deep learning,” Neural Computing and Applications, vol. 31, no. 12, pp. 9349–9361, Dec. 2019.

[9] S. Garg, A. Saxena, and R. Gupta, “Yoga pose classification: A CNN and MediaPipe inspired deep learning approach for real-world application,” Journal of Ambient Intelligence and Humanized Comput-ing, vol. 14, no. 12, pp. 16 551–16 562, Dec. 2023.

[10] A. Bera, M. Nasipuri, O. Krejcar, and D. Bhattacharjee, “Fine-Grained Sports, Yoga, and Dance Pos-tures Recognition: A Benchmark Analysis,” IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1–13, 2023.

[11] R. R. Subramanian and V. Govindaraj, “HARNet: Design and evaluation of a deep genetic algorithm for recognizing yoga postures,” Signal, Image and Video Processing, vol. 18, no. S1, pp. 553–564, Aug. 2024.

[12] Y. Zhou, S. Dong, H. Sheng, and W. Ke, “YED-Net: Yoga Exercise Dynamics Monitoring with YOLOv11-ECA-Enhanced Detection and DeepSORT Tracking,” Applied Sciences, vol. 15, no. 13, p. 7354, Jun. 2025.

[13] J. Palanimeera and K. Ponmozhi, “Classification of yoga pose using machine learning techniques,” Materials Today: Proceedings, vol. 37, pp. 2930–2933, 2021.

[14] S. Garg, A. Saxena, and R. Gupta, “Yoga pose classification: A CNN and MediaPipe inspired deep learning approach for real-world application,” Journal of Ambient Intelligence and Humanized Comput-ing, vol. 14, no. 12, pp. 16 551–16 562, Dec. 2023.

[15] D. Swain, S. Satapathy, B. Acharya, M. Shukla, V. C. Gerogiannis, A. Kanavos, and D. Giakovis, “Deep Learning Models for Yoga Pose Monitoring,” Algorithms, vol. 15, no. 11, p. 403, Oct. 2022.

[16] D. Deepa, R. Velumani, and S. Selvaraj, “Yoga Pose Estimation along with Human Posture Detection using Deep Learning Approach,” Applied and Computational Engineering, vol. 2, no. 1, pp. 993–1000, Mar. 2023.

[17] R. Gadhvi, P. Desai, and Siddharth, “PosePilot: An Edge-AI Solution for Posture Correction in Physical Exercises,” May 2025.

[18] R. Jadhav, V. Ligde, R. Malpani, P. Mane, and S. Borkar, “Aasna: Kinematic Yoga Posture Detection And Correction System Using CNN,” ITM Web of Conferences, vol. 56, p. 05007, 2023.

[19] S. S and H. Kumar, “Recognizing Yoga Pose using Deep Learning,” International Journal of Innovative Science and Research Technology (IJISRT), pp. 1124–1128, Jul. 2024.

Share your valuable work from Social Media Buttons