FEDS -- Filtered Edit Distance Surrogate

TL;DR

This paper introduces a robust surrogate for edit distance in scene text recognition, utilizing self-paced learning to filter difficult examples, leading to significant improvements in recognition accuracy across multiple datasets.

Contribution

It presents a novel end-to-end training procedure using a learned surrogate of edit distance with self-paced filtering, enhancing scene text recognition performance.

Findings

Average 11.2% improvement in total edit distance

9.5% reduction in error rate on accuracy

Effective on diverse challenging datasets

Abstract

This paper proposes a procedure to train a scene text recognition model using a robust learned surrogate of edit distance. The proposed method borrows from self-paced learning and filters out the training examples that are hard for the surrogate. The filtering is performed by judging the quality of the approximation, using a ramp function, enabling end-to-end training. Following the literature, the experiments are conducted in a post-tuning setup, where a trained scene text recognition model is tuned using the learned surrogate of edit distance. The efficacy is demonstrated by improvements on various challenging scene text datasets such as IIIT-5K, SVT, ICDAR, SVTP, and CUTE. The proposed method provides an average improvement of $11.2 \%$ on total edit distance and an error reduction of $9.5\%$ on accuracy.