ThermalDiff: A diffusion architecture for thermal image synthesis

Tayeba Qazi, Brejesh Lall, Prerana Mukherjee

Abstract: Highlights•The diffusion model serves as an effective framework for estimating thermal images from visible-spectrum inputs.•Our conditional variant of the diffusion model effectively learns both structural and temperature variations.•Evaluated on three benchmark datasets, our method achieves state-of-the-art performance.•Pedestrian detection performance improves by approximately 8% using our synthesized thermal data.

External IDs:dblp:journals/jvcir/QaziLM25