Generative Shape Models: Joint Text Recognition and Segmentation with Very Little Training Data

Autor:	Lou, Xinghua, Kansky, Ken, Lehrach, Wolfgang, Laan, CC, Marthi, Bhaskara, Phoenix, D. Scott, George, Dileep
Rok vydání:	2016
Předmět:	Computer Science - Computer Vision and Pattern Recognition
Zdroj:	Advances in Neural Information Processing Systems 2016
Druh dokumentu:	Working Paper
Popis:	We demonstrate that a generative model for object shapes can achieve state of the art results on challenging scene text recognition tasks, and with orders of magnitude fewer training images than required for competing discriminative methods. In addition to transcribing text from challenging images, our method performs fine-grained instance segmentation of characters. We show that our model is more robust to both affine transformations and non-affine deformations compared to previous approaches.
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/1611.02788 Zobrazit plný text záznamu View this record from Arxiv