MediaPipe Based Fitness Workout Pose Estimation Approach towards Physical Exercise Detection and Calorie Burnt Estimation on Tracking key Body Landmark Movements
Year: 2025
Type: article
Abstract: Nowadays, lifestyle changes and mental stress on the peoples make them obese with extra fat deposited in thebody and it leads to various health complication. In order to burn extra fat deposited in the body, gym workoutbecome important which burn extra and unwanted calories through different type of workouts. Especiallyadvancement of technologies helps to track the fitness of the health in real time. Technologies uses devices such assensor like accelerometers and altimeter to gather movement of the body and it is processed using machine learningand deep learning model to determine the calories burnt. However these model are highly expensive and timeconsuming and it requires receiver and transmit unit which increase the device size. Further to handle thesechallenges, device free method has been employed to process the human movement gathered through camera in realtime or through recorded video. Tracking fully body postures of the workout in real time process the human bodymovement through MediaPipe which is considered as pose estimation approach. MediaPipe perform machinelearning task like pose estimation which is able to extract the human movement like arm gestures, leg lifts , squatsand jumps on relative difference between body parts as feature points. Feature points were processed to identify themovement of the body landmarks related to the specified exercise and to detect motion pattern corresponds tospecific exercise. Based on detected human body movement, time series data is generated. These data is processedfurther to determine the calories burnt and to track the progress to provide instant feedback to enhance workoutperformance and its routines. Experimental analysis of the proposed model proves that pose detection based fullybody workout represents immense potential in generating interest in the fitness enthusiasts. Further it helps toperform exercise with proper form on correcting the movement patterns to eliminate the risk of injuries.Furthermore, particular model provides increased accuracy and efficiency compared to other existing models.
Author: K.Karthikeyan, V.Senthilkumar
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W6912216037
FWCI:
Cites:
Cited by:
Related to: