Article: Ensemble of RNN classifiers for activity detection using smartphone and supporting nodes

The dataset was provided for classification of activity using sensors placed in four places: waist, chest, arm and leg.

The dataset contains eight classes:

• walking in normal paste, where legs and arms are moving without any luggage,
• jogging in moderate paste, where limb movement is faster, and the body is moving more dynamically,
• squats, where a person is performing squats in place and arms holding a balance,
• jump, where a person jumps in place as a part of aerobic exercises,
• laying, where a person is lying on his back, and major movement corresponds to chest breathing,
• arms swing, where the arms are moving during the exercise or housework,
• siting, where the person sits at the desk and performs simple writing (office) activities.
• standing in place, where only natural body balance can be registered.

The repository can be found in: https://github.com/drOneMB/activity-detection