This project involves the development of a 3D skeletal data file converter and graphical interface that takes Human Action Recognition (HAR) data (mainly available in csv files) and converts it json files.
Human Action Recognition (HAR) consists in automatically recognising motions performed by people from sensors that can detect and track the users’ skeleton. These provide the positions and the orientations of the body joints over a time frame. Most of the existing and available HAR datasets such as H.3.6m or UCI store data in one or several csv files.
We have a tool able to animate a skeleton from a json file. So, the objective of this project is to create an interface able to load any csv file and convert the skeleton data into our json format.
This project is suitable for a FYP