InstaCities1M
收藏资源简介:
InstaCities1M由与世界上10个人口最多的英语城市之一相关的Instagram图像形成。每个城市有100K个图像,总共有1M个图像,分为800K个训练图像,50K验证图像和150K个测试图像。这个数据集的兴趣是: 首先是由最近的社交媒体数据形成的。与图像关联的文本是描述和照片加载器编写的主题标签。因此,这是一种免费的可用数据,可以从中学习。其次,每个图像都与一个城市相关联,因此我们可以使用这种弱标记的图像类别来评估我们的实验。所有图像的大小都调整为300x300像素。
InstaCities1M is constructed from Instagram images associated with one of the 10 most populous English-speaking cities globally. Each city has 100K images, totaling 1 million images overall, which are split into 800K training images, 50K validation images, and 150K test images. The notable characteristics of this dataset are as follows: Firstly, it is compiled from recent social media data. The text paired with each image includes descriptions and hashtags written by image uploaders, making this freely accessible data suitable for model learning. Secondly, every image is linked to a specific city, enabling us to use these weakly labeled image categories to evaluate our experimental studies. All images have been resized to a resolution of 300×300 pixels.




