Anime character segmentation with UNet

Classes : [ background, hair, eye, mouth, face, skin, clothes ]

Model

In: 3x512x512 -> Out: 7x512x512

It uses pretrained mobilenet_v2 as encoder

References

[1] Deep Learning Project — Drawing Anime Face with Simple Segmentation Mask link

[2] pit-ray/Anime-Semantic-Segmentation-GAN link