Chenming Li Shiguang Liu PredVSD: Video saliency prediction based on conditional diffusion model. 113820 2025 324 Knowl. Based Syst. https://doi.org/10.1016/j.knosys.2025.113820 db/journals/kbs/kbs324.html#LiL25 streams/journals/kbs