StreetStyle
收藏资源简介:
每天有数十亿张照片上传到照片共享服务。这些图像包含有关人们如何在世界各地生活的信息。在本文中,我们利用这些丰富的数据来了解全球时尚和风格趋势。我们提出了一个大规模视觉发现框架,分析世界各地数百万张图像中的服装和时尚,跨越数年。我们引入了一个带有服装属性注释的大规模人物照片数据集,并使用该数据集通过深度学习训练属性分类器。我们还提出了一种发现视觉一致的风格集群的方法,这些集群在这个庞大的数据集中捕获有用的视觉相关性。有了这些工具,我们分析了数百万张照片以获得视觉洞察力,对全球和每个城市的时尚选择和时空趋势进行了首创的分析。
Billions of photos are uploaded to photo-sharing services daily. These images hold information about how people live their lives around the world. In this paper, we leverage this rich dataset to investigate global fashion and style trends. We propose a large-scale visual discovery framework that analyzes clothing and fashion in millions of images across the globe over multiple years. We introduce a large-scale dataset of human portraits annotated with clothing attributes, and utilize this dataset to train attribute classifiers via deep learning. We also present a method for discovering visually consistent style clusters that capture meaningful visual correlations within this massive dataset. Armed with these tools, we analyze millions of photos to derive visual insights, conducting the first-of-its-kind analysis of fashion preferences and spatiotemporal trends at both global and city-specific levels.




