CEKD: Cross-Modal Edge-Privileged Knowledge Distillation for Semantic Scene Understanding Using Only Thermal Images | IEEE Journals & Magazine | IEEE Xplore