๐ Are you at #ICCV2025?
Check out Poster #125 where we present ScanDiff:
โ
Generates diverse, realistic human scanpaths
โ
Uses diffusion + ViTs + optional text conditioning
โ
Beats SOTA in free-viewing & task-driven tasks
Come talk with us!
#DiffusionModels #GazePrediction #ICCV2025
0
0
0
0