Image Dataset of L10n Bugs in Open Source Projects

This dataset contains 6,410 manually curated images of localization (L10n) bugs collected from public GitHub issues in open source software projects. The dataset was developed through a two-stage process consisting of automated issue collection and manual validation of visual evidence. Each valid image was categorized according to the localization bug type, software platform, and language involved. The collection covers 76 languages and multiple software platforms, including Web, Windows, Android, iOS, Linux, macOS, and digital games. The accompanying metadata file provides detailed information about each image, including the original GitHub issue, repository, creation date, repository license, search terms used during collection, bug category, platform, language, and image source URL. The dataset was created as part of the Master's research project "Construção e Caracterização de um Dataset de Imagens de Bugs de Localização (L10n) em Projetos Open Source", developed in the Master's Program in Software Engineering at the Federal University of Pernambuco (Universidade Federal de Pernambuco – UFPE). This dataset aims to support empirical studies on software localization, software quality assurance, bug characterization, and the development of automated approaches for localization bug detection using computer vision and multimodal machine learning techniques.