Fashion Landmark Detection

时尚地标检测基准评估时尚地标检测的性能。这是 DeepFashion 的一个大子集，具有多样且大的姿势/放大变化。它包含 123,016 个衣服图像；每个图像的 8 个时尚地标（位置和可见性）；每个图像还通过边界框、服装类型和变化类型进行注释。

The Fashion Landmark Detection Benchmark evaluates the performance of fashion landmark detection. It is a large subset of DeepFashion, featuring diverse and substantial variations in pose and magnification. The dataset includes 123,016 clothing images, with each image annotated with 8 fashion landmarks along with their respective locations and visibility status. Additionally, each image is also annotated with bounding boxes, clothing categories and variation types.

搜集汇总

背景与挑战

背景概述

该数据集是DeepFashion的一个大规模子集，专注于时尚地标检测，包含123,016张衣服图像。每张图像均标注了8个时尚地标的位置和可见性，同时提供了边界框、服装类型和变化类型等丰富注释，由香港中文大学于2016年发布。