Abstract
The growing field of HAR sees many applications from remote health monitoring for the elderly to smart homes that adapt to our daily routines.
In our work, we investigated the performance of three state-of-the-art spatio-temporal graph neural networks against traditional deep learning models, setting clear performance benchmarks. We also extended this to human activity prediction, looking not just at what someone is doing, but what they’re likely to do next.
Finally, we built a 3D virtual environment that lets us visualise sensor data in real time. This environment can be used for data augmentation- the generation of new training data to help future researchers build more accurate and robust AI systems.
Videos 1
Watch presentations, demos, and related content
Video summary of the work conducted during our 3DHAR-AI project.
Like, comment, and subscribe on YouTube to support the creator!
Video summary of the work conducted during our 3DHAR-AI project.
Gallery 1
Explore the visual story of this exhibit
Title Image