• Industrial inspection: X-ray object recognition and nondestructive testing, including multi-view detection and public benchmarks such as GDXray.
• Machine learning for scientific and biomedical data: learning under data scarcity and class imbalance, from medical data to phage–host interaction prediction and variable star classification.
• Applied perception systems: vision-language models for quality assessment and tactile object recognition in robotic grippers.
• (Emerging) self-supervised learning and representation learning: learning representations from unlabeled data, with ongoing work on object counting and re-identification.