ViTMatte: Boosting Image Matting with Pretrained Plain Vision Transformers
Open paper
Benchmark
Model
Rank
Results
image-matting-on-composition-1k
ViTMatte
#3
MSE: 3.0
SAD: 20.33
Conn: 14.78
Grad: 6.74
image-matting-on-distinctions-646
ViTMatte
#2
SAD: 17.05
MSE: 0.0015
Grad: 7.03
Conn: 12.95
Trimap: √
Rank counts only results with a code link.