Yoga, an ancient discipline that originated 5000 years ago in India, provides a holistic approach to harmonizing the body and mind through features such as asanas, meditation, and breathing methods. With the global increase in stress, yoga has grown in popularity, encouraging a variety of learning techniques. Self-practice, the favored method, has difficulties discovering pose inaccuracies. In our suggested system, users choose a position, upload a photo, and get feedback by comparing their posture to an expert’s. Calculating differences in joint angles allows for independent pose refining, which bridges gaps in self-learning. Approach improves personalization, leading to a more effective yoga experience. Recognizing the fast-paced nature of modern lifestyles, this article addresses the demand for home-based fitness by investigating machine learning and deep learning algorithms for correct yoga posture classification.The study delves into position estimation and keypoint recognition technologies, laying the groundwork for a self-instruction exercise system that allows anyone to learn and practice yoga confidently even without a specialized instructor.