Michal Mackiewicz

This PhD project at the University of East Anglia focuses on developing advanced multispectral deep learning algorithms for the classification of marine litter. Marine litter poses a significant threat to ocean health and livelihoods, necessitating scalable and automated data collection and analysis methods. The project builds on existing visible light (VL) datasets and deep learning (DL) algorithms developed by Cefas, aiming to enhance classification accuracy through the integration of multispectral imagery. The student will collect new multispectral data using a state-of-the-art lab setup and train DL models to classify marine litter types, ensuring robustness to varying illumination conditions and sensor configurations. The research will address the physics of multispectral image formation, considering sensor spectral sensitivities, daylight spectrum variations, and material reflectance spectra. Key objectives include expanding the VL dataset to multispectral imaging, designing and evaluating robust DL models, and implementing device-independent representations. The project will also explore domain adaptation and transfer learning to generalize models across different imaging devices, including future sensors. Training will be provided in imaging principles, lab measurement, computer vision, ArcGIS, fieldwork, and UAV operation. The student will be based in the Colour & Imaging Lab at the School of Computing Sciences and collaborate with scientists and engineers at Cefas. Applicants should have a strong background or interest in environmental monitoring, AI, computer vision, or multispectral imaging, and hold at least a UK equivalent Bachelors (Honours) 2:1 degree in a numerate discipline. English language proficiency is required. Funding is available through ARIES studentships, covering fees, a maintenance stipend (£20,780 p.a. for 2025/26), and a research training and support grant. International applicants may have the fee difference waived, but relocation and visa costs are not covered. The position is full-time and starts on 1 October 2026.