2026-10-13T00:24:47Z https://www.scidb.cn/oai

10.57760/sciencedb.j00001.01639 2026-04-16T10:17:54Z

2026-04-16 International Phonetic Alphabet Image-Text Alignment Dataset doi:10.57760/sciencedb.j00001.01639 en The dataset consists of three annotation files and one image directory. (1) train_label.json, valid_label.json and test_label.json are annotation files, which are divided into training, validation and test sets with a ratio of 7:2:1, respectively. Each file contains aligned image–text pairs, with data volumes of 9.4 MB, 3.1 MB and 2.1 MB, respectively; (2) images is an image directory, which contains all IPA images corresponding to the image paths in the annotation files, with a total data volume of 409.6 MB. IPA image dataset; IPA recognition; deep learning; YOLO network Long Congjun Wu Shuaizhen EMBARGO https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank