I am Dr Phil Birch, and I am a Reader (Associate Prof.) at the University of Sussex, Department of Engineering and Design and a member of the Industrial Informatics and Signal Processing group. This website outlines my research interests and lists some useful resources.
My main research area crosses the boundary of optics and image processing. I have interests in:
- Human and Vehicle Tracking
- Medical Image Processing
- Fluorescence Life Imaging Microscopy
- Optical Signal Processing and correlation
Examples of my research include:
- Drone (UAV) detection. Due to the small size of the drones, this turns out to rather difficult. Standard object detection methods, such as YOLO, don’t do well when the object is only a few pixels across.
- People tracking — or multi-object tracking. Tracking people in crowds is tricky. Linear predictors, such as a Kalman filter, work reasonable well when people walk in a straight line. They fail, though, when the crowd gets too busy and people have to move around each other. We have been investigating using deep learning to learn these “social interactions”, and predict how people will move in a crowd. The ETTrack video shows this working on the DanceTrack Dataset.
- One of the major problems with medical computer vision task, e.g., cancer detection from MRI/CT, is the lack of training data. We are looking at how generative models, such as diffusion models, can be used to augment datasets with synthetic data. This could be then applied to other areas of computer vision.
I am looking for PhD students that are interested in optics, optical simulations, object tracking, and deep learning. If you are interested, please get in contact.
Publications
My publications can found on the University’s archive server
http://orcid.org/0000-0002-7740-9379
Contact
I can be contacted via this page.