Human Activity Recognition: predicting dumbbell technique success from wearables.
People regularly quantify how much of a particular activity they do with wearable devices, but they rarely quantify how well they do it.
This project seeks to determine the potential for real-time user feedback from model-based assessment of the quality, not quantity, of exercise technique.
To this end, six participants were asked to perform dumbbell lifts correctly and incorrectly in five different ways while accelerometers on the belt, forearm, arm, and dumbbell recorded their movements.
More information is available here (see the section on the Weight Lifting Exercise Dataset).
The data for this project kindly provided by Groupware@LES
Download the training data (csv, 12 MB)
Download the test data (csv, 15 KB)
Analyses were peformed using R. Reporting written in Rmarkdown format and rendered in HTML using knitr.
View the published report at https://msinjin.github.io/dumbbell/.
Velloso, E.; Bulling, A.; Gellersen, H.; Ugulino, W.; Fuks, H. Qualitative Activity Recognition of Weight Lifting Exercises. Proceedings of 4th International Conference in Cooperation with SIGCHI (Augmented Human '13) . Stuttgart, Germany: ACM SIGCHI, 2013.