use the mouse or number keys 1–8 to switch between images
1: input
2: ground truth shadow-free
3: shadow mask
4: reconstructed matte (ours)
5: our output
6: Guo et al. 2012
7: Guo et al. 2012 (automatic detection)