At the CV4Ecology Workshop at ECCV 2026, Christoph presented two of our recent papers together with Hugo Markoff, showcasing new approaches to AI-based wildlife monitoring from aerial RGB and thermal video.
The first paper, When One Modality Is Not Enough: Multimodal Sex and Life-Stage Classification of Red Deer from Aerial RGB/Thermal Video, explores how combining RGB and thermal imagery can improve the classification of red deer by sex and life stage. By fusing complementary information from both sensors, the approach can handle cases where either modality alone is insufficient.
The second paper, Oh Deer, How Should I Handle This? Seasonal Priors for Selective Wildlife Annotation and Classification, investigates how seasonal biological changes can be incorporated into wildlife annotation and classification. The work shows how ecological knowledge and uncertainty-aware prediction can make automated monitoring more reliable.
Together, the two papers highlight the value of combining multimodal sensing, computer vision, and ecological knowledge for more robust and informative wildlife monitoring.