CTW1500
收藏资源简介:
CTW1500数据集是由华南理工大学电子信息工程学院创建的一个新型曲线文本数据集,专注于提升自然场景中曲线文本的检测能力。该数据集包含1500张图像,总计超过10000个文本标注,每张图像至少包含一个曲线文本。数据来源多样,包括互联网、图像库及手机拍摄等,涵盖多种语言和文本方向。创建过程中,研究团队采用了14点多边形标注方法,有效减少了人工标注的工作量。CTW1500数据集的应用领域主要集中在曲线文本检测,旨在解决现有数据集中曲线文本标注不足的问题,推动曲线文本检测技术的发展。
CTW1500 dataset is a novel curved text dataset created by the School of Electronic and Information Engineering, South China University of Technology, which focuses on enhancing the detection performance of curved text in natural scenes. It consists of 1500 images with a total of over 10,000 text annotations, and each image contains at least one curved text instance. The dataset has diverse sources, including the Internet, image libraries, mobile phone photography and so on, covering multiple languages and text orientations. During the dataset construction, the research team adopted a 14-point polygonal annotation method, which effectively reduced the workload of manual annotation. The CTW1500 dataset is mainly applied to curved text detection, aiming to address the shortage of curved text annotations in existing datasets and promote the development of curved text detection technologies.

- 1Detecting Curve Text in the Wild: New Dataset and New Solution华南理工大学电子信息工程学院 · 2017年



