Computer vision enables AI models to understand, analyse, and now even generate, complex image and video data.” Ivor Simpson
Lecturer in Artificial Intelligence
Image and video data are ubiquitous in our digital world. Ranging from ordinary photographs and social media videos, to video streams from autonomous vehicles, satellite imaging and magnetic resonance images of the human brain. Computer vision provides us with a flexible set of approaches and tools to analyse and make predictions from such data. The AI research group has wide ranging expertise in tackling these problems using a combination of deep convolutional neural networks, statistical modelling and traditional image processing techniques. We work on a wide range of applications including: aerial image segmentation; image editing for creative applications; video analysis; and neuroimage analysis for biomarker discovery.
We make extensive use of advanced machine learning approaches to enable weakly-/semi- supervised learning, estimates of prediction uncertainty, and fair data representations. We also have a strong focus on deep generative models such as variational auto-encoders (VAEs) and generative adversarial networks (GANs), which can be used in many applications.
For further details of some of our Computer Vision research projects please see the Predictive Analytics Laboratory website.