Can Generative Adversarial Networks Teach Themselves Text Segmentation?

In the information age in which we live, text segmentation from scene images is a vital prerequisite task used in many text understanding applications. Text segmentation is a difficult problem because of the potentially vast variation in text and scene landscape. Moreover, systems that learn to perform text segmentation usually need non-trivial annotation efforts. We present in this work a novel unsupervised method to segment text at the pixel-level from scene images. The model we propose, which relies on generative adversarial neural networks, segments text intelligently; and does not therefore need to associate the scene image that contains the text to the ground-truth of the text. The main advantage is thus skipping the need to obtain the pixel-level annotation dataset, which is normally required in training powerful text segmentation models. The results are promising, and to the best of our knowledge, constitute the first step towards reliable unsupervised text segmentation. Our work opens a new research path in unsupervised text segmentation and poses many research questions with a lot of trends available for further improvement.