Annotated Infant Fundus Images for Object Detection of Key Structures (Optic Disk and Macula)

The data set contains fundus images for object detection research. Some images have clipped corners (patient info watermark is printed in this area) to ensure they are de-identified. All images are annotated by one senior pediatric ophthalmologist in SEH (Shenzhen Eye Hospital). An image annotation tool is developed based on the "VGG Image Annotator (VIA)" tool from the University of Oxford.The tool is included in this repository. The optic disk and macula in each fundus image are annotated by rectangular bounding boxes. This repository also contains related materials for our case study on this data set.

本数据集包含用于目标检测研究的眼底图像。部分图像的边角已被裁切（该区域印有患者信息水印），以确保数据完成匿名化处理。所有图像均由深圳眼科医院（Shenzhen Eye Hospital, SEH）的一名资深小儿眼科医师标注。本数据集配套的图像标注工具基于牛津大学研发的“VGG图像标注器（VGG Image Annotator, VIA）”定制开发，该工具已包含在本代码库中。每张眼底图像中的视盘（optic disk）与黄斑（macula）均通过矩形边界框进行标注。本代码库同时包含针对本数据集开展案例研究的相关资料。