Pose Estimation and Virtual Gym Assistant Using MediaPipe and Machine Learning

This research paper demonstrates a virtual fitness trainer, using Google's Mediapipe library, which was made for various multimodal machine learning and deep learning pipelines. By monitoring the user's every step with posture estimation algorithms, the suggested system can provide useful feedback and insights as they work out. Mediapipe's deep algorithms and posture estimation module are utilized to develop this novel system, which captures user movements by identifying unique body landmarks for each exercise. The system also keeps track of the total amount of reps quite precisely. The system determines the user's performance by computing angles and landmarks and feeding this data into a machine-learning model, which then classifies the correct posture and the number of repetitions of the exercise. The system uses many fitness datasets in addition to user-supplied data for thorough training and assessment. A detailed comparison of machine learning models like Logistic Regression, Support Vector Machine, Naive Bayes, Decision Tree, and Artificial Neural Network is included in this research. The findings prove that our suggested method provides accurate and individualized feedback, which improves exercise performance and lessens the likelihood of damage.