1 paper
Shilin Hu, Jingyi Xu, Sagnik Das +2
Shadows encode rich information about scene geometry and illumination, yet existing methods either predict a unified shadow mask or overlook attached shadows entirely. We address t…