Rooms with Text: A Dataset for Overlaying Text Detection

TL;DR

This paper introduces a new dataset of interior room images with overlaying text and proposes a baseline detection method that achieves high accuracy, facilitating research in scene text detection within indoor environments.

Contribution

The paper provides a novel dataset of 4836 annotated interior images with overlaying text and a baseline detection approach leveraging character region awareness.

Findings

Achieved 0.95 F1 score in text detection

False positive rate of 0.02

False negative rate of 0.06

Abstract

In this paper, we introduce a new dataset of room interior pictures with overlaying and scene text, totalling to 4836 annotated images in 25 product categories. We provide details on the collection and annotation process of our dataset, and analyze its statistics. Furthermore, we propose a baseline method for overlaying text detection, that leverages the character region-aware text detection framework to guide the classification model. We validate our approach and show its efficiency in terms of binary classification metrics, reaching the final performance of 0.95 F1 score, with false positive and false negative rates of 0.02 and 0.06 correspondingly.