Tear Film Multitask (TFM) Dataset
收藏资源简介:
TFM数据集是一个用于多任务泪膜分析的综合性公开数据集,包含从15个高分辨率视频中提取的6,247张图像,每张图像都被精心标注了三个互补的视觉任务:图像分类(将帧分为'清晰'、'闭合'、'破裂'和'模糊')、目标检测(定位普拉西多环和瞳孔区域)和像素级语义分割(勾画泪膜破裂区域)。这种多任务标注方案为开发和应用全面的泪膜分析模型提供了丰富的资源。
The TFM Dataset is a comprehensive public dataset for multi-task tear film analysis. It contains 6,247 images extracted from 15 high-resolution videos. Each image has been meticulously annotated with three complementary visual tasks: image classification (classifying frames into "clear", "closed", "ruptured" and "blurred"), object detection (localizing the Placido ring and pupil regions), and pixel-level semantic segmentation (outlining the tear film rupture regions). This multi-task annotation scheme provides a rich resource for the development and deployment of comprehensive tear film analysis models.




