ReLaText: Exploiting visual relationships for arbitrary-shaped scene text detection with graph convolutional networks

Chixiang Ma, Lei Sun, Zhuoyao Zhong, Qiang Huo