Commercial water sports centres such as Cliff Lakes, Wild Shore, Foxlake Adventures, and others have been popping up throughout the UK since the early 2000s. Unlike traditional water sports centres, these companies specialise in extreme water sports and high-throughput activities.

But with all adventurous activities, there is a degree of risk, this risk being what attracts consumers to the activity. But from a company’s perspective, they have both a social and legal responsibility to mitigate this risk. These sorts of activities hold a greater risk, as not only are the activities inherently risky, but also the environment creates additional risks.

This project aims to improve safety for consumers of water sports centres. It creates a proof of concept application that provides activity centre instructors with a single multi-application tool to increase safety. Focusing on the two largest challenges identified by market research.

The application takes a live video stream and analyses each frame individually. For its primary function of counting participants entering and exiting it utilises YOLO’s v11 person detection along with object tracking and trajectory analysis to identify people that cross the boundary.

With a further PPE detection tool that utilises an additional model to identify people who are missing buoyancy aids. This model utilises a custom-trained YOLO model that has been trained on a custom-labelled dataset. As well as combining a multi-detection required pipeline for improved accuracy.

Data analysis is provided through the application’s graphical user interface, built in ASP.NET. The application can control the monitoring and pipeline, as well as display core statistics and alerts for when no PPE is detected.

The final application is a proof of concept, which would require further development was it to be used commercially. The project also created an artefact of a BA-labelled dataset.