Edit Probability for Scene Text Recognition

Abstract

We consider the scene text recognition problem under the attention-based encoder-decoder framework, which is the state of the art. The existing methods usually employ a frame-wise maximal likelihood loss to optimize the models. When we train the model, the misalignment between the ground truth strings and the attention's output sequences of probability distribution, which is caused by missing or superfluous characters, will confuse and mislead the training process, and consequently make the training costly and degrade the recognition accuracy. To handle this problem, we propose a novel method called edit probability (EP) for scene text recognition. EP tries to effectively estimate the probability of generating a string from the output sequence of probability distribution conditioned on the input image, while considering the possible occurrences of missing/superfluous characters. The advantage lies in that the training process can focus on the missing, superfluous and unrecognized characters, and thus the impact of the misalignment problem can be alleviated or even overcome. We conduct extensive experiments on standard benchmarks, including the IIIT-5K, Street View Text and ICDAR datasets. Experimental results show that the EP can substantially boost scene text recognition performance.

References in corpus (8)

• Sequence to Sequence Learning with Neural Networks
• ADADELTA: An Adaptive Learning Rate Method
• Caffe: Convolutional Architecture for Fast Feature Embedding
• Synthetic Data and Artificial Neural Networks for Natural Scene Text Recognition
• Synthetic Data for Text Localisation in Natural Images
• End-to-End Text Recognition with Hybrid HMM Maxout Models
• Deep Structured Output Learning for Unconstrained Text Recognition
• Recursive Recurrent Nets with Attention Modeling for OCR in the Wild

Cited by in corpus (8)

• Scene Text Recognition from Two-Dimensional Perspective
• Text Detection and Recognition in the Wild: A Review
• Recurrent Calibration Network for Irregular Text Recognition
• Generating Text Sequence Images for Recognition
• Primitive Representation Learning for Scene Text Recognition
• SAFE: Scale Aware Feature Encoder for Scene Text Recognition
• A Compositional Textual Model for Recognition of Imperfect Word Images
• Exploring Font-independent Features for Scene Text Recognition