DeepFashion2
收藏资源简介:
DeepFashion2是由香港中文大学创建的大型时尚图像数据集,包含491,000张图像和801,000个服装项目,每个项目都有丰富的标注,如风格、尺度、视角、遮挡、边界框、密集地标和掩码。数据集通过控制尺度、遮挡、缩放和视角的变量来收集,以创建一个具有挑战性的基准。DeepFashion2适用于多种时尚理解任务,包括服装检测、姿态估计、分割和检索,旨在解决时尚图像分析中的实际问题。
DeepFashion2 is a large-scale fashion image dataset developed by The Chinese University of Hong Kong. It contains 491,000 images and 801,000 clothing items, each with rich annotations including style, scale, viewpoint, occlusion, bounding boxes, dense landmarks and masks. The dataset is collected by controlling variables such as scale, occlusion, zoom and viewpoint to establish a challenging benchmark. DeepFashion2 supports a variety of fashion understanding tasks, including clothing detection, pose estimation, segmentation and retrieval, and aims to solve practical problems in fashion image analysis.




