Open AccessAUTO_FIT: WORKOUT TRACKING USING POSE-ESTIMATION AND DNN
Author(s) -
Nitesh Sonwani,
Aryan Pegwar
Publication year - 2020
Publication title -
international journal of engineering applied science and technology
Language(s) - English
Resource type - Journals
ISSN - 2455-2143
DOI - 10.33564/ijeast.2020.v05i01.024
Subject(s) - pose , artificial intelligence , computer science , tracking (education) , computer vision , estimation , engineering , psychology , pedagogy , systems engineering
Lack of physical fitness increases the risk of adverse health conditions including coronary heart diseases, high blood pressure, stroke, metabolic syndrome, type 2 diabetes which leads to a decrease in the life expectancy of humans. In our work, we have introduced Auto_fit, an application that suggests the workouts and tracks it. Auto_fit uses Postnet for doing pose estimation to find 17 body keypoints followed by using the DNN classifier to identify the state of exercise and then counts the repetitions performed. We collected the videos of trained professionals performing the exercise and then used it to train Auto_fit. Auto_fit takes live video feed and counts the repetitions of exercise performed. It works on two common exercises and can also be run on low single-board computers like Raspberry pi. Auto_fit helps in improving physical fitness and thus enables a person to live a longer and healthier life. Keywords— DNN, Posenet, Raspberry pi, Pose estimation,