1st Place Solution to ECCV 2022 Challenge on Out of Vocabulary Scene Text Understanding: End-to-End Recognition of Out of Vocabulary Words

TL;DR

This paper presents a top-performing end-to-end scene text recognition model for out-of-vocabulary words, leveraging oCLIP, achieving state-of-the-art results in the ECCV 2022 OOV challenge.

Contribution

The authors introduce an oCLIP-based approach specifically designed for recognizing out-of-vocabulary words in natural scene images, setting a new benchmark in the field.

Findings

Achieved 28.59% h-mean score, ranking 1st in the ECCV 2022 OOV challenge.

Demonstrated effectiveness of oCLIP in out-of-vocabulary scene text recognition.

Outperformed existing methods in end-to-end OOV word recognition.

Abstract

Scene text recognition has attracted increasing interest in recent years due to its wide range of applications in multilingual translation, autonomous driving, etc. In this report, we describe our solution to the Out of Vocabulary Scene Text Understanding (OOV-ST) Challenge, which aims to extract out-of-vocabulary (OOV) words from natural scene images. Our oCLIP-based model achieves 28.59\% in h-mean which ranks 1st in end-to-end OOV word recognition track of OOV Challenge in ECCV2022 TiE Workshop.