This project will develop computational imaging methods to reconstruct and track the behavior of freely moving fruit flies across long, multiview video recordings. The project will use conditional Gaussian splatting-based 4D reconstruction to identify behavioral phenotypes and connect them with neural activity and physiological state, creating tools with potential applications across neuroscience and ethology.
2026
Temporally scalable 4D reconstruction and neurobehavioral phenotyping of freely moving animals
(top) Simplified schematic of the forward ray-based model of our multiview computational imaging setup. (bottom) Simplified overview of our spatiotemporally scalable 4D reconstruction algorithm, based on self-supervised conditional 4D Gaussian splatting.
Other Researchers
Kevin Zhou, Assistant Professor, Biomedical Engineering
Ravi Allada, Theophile Raphael M.D. Collegiate Professor of Neurosciences