oneTrack was built at JISC’s CAN Conference, built around our own experiences of lectures, and seeing students disengage. We thought to work around JISC’s intelligent campus strand, but creating a solution to allow staff to understand where students are engaging and why engagement could be dropping across lectures.
This conference was about the Change Agents Network, a network of staff and students working in partnership to support curriculum enhancement and innovation across higher and further education globally. We decided to target our entry to this to create a product that would help staff and students alike to improve the lecture experience.
The application consisted of two elements, both built-in cross-platform .NET Core 2.1, the client application and the server processing side, with the client running in lecture theatres and the server running remotely (likely somewhere in the cloud).
The client side captures images from a camera pointing towards the audience and sends it to Azure’s face recognition API to get a JSON file back of statistics of people in the picture, the frame is marked up with red squares for where people are located through openCV and passed to the server side.
The server application then calculates statistics across the session, to calculate the averages for attentiveness, engagement and score the session before storing it into the database. The frontend then uses ChartJS and some friendly graphs to portray this information to the user.
If this was being used in production, we’d also link this to Panopto (or other lecture capture software) to overlay attentiveness over the stream to allow the member of staff to understand which parts of the session resulted in increased and reduced engagement.