method: HappyCCL2018-01-16

Authors: Wenhua Cheng; Anbang Yao; Jianguo Li; Dongqi Cai

Description: FCN based. Only 1.0k images and Vgg16 are used.