Diversity in AI

Daily objects around the world dataset

In collaboration with Gapminder

Humans in the Loop is happy to publish open access annotations for Gapminder’s Dollar Street dataset. 

The Dollar Street project aims to show people around the world and how they really live. With more than 27,000 images taken across 50 countries, the dataset covers a variety of daily objects and scenes. 

Humans in the Loop has performed bounding box and image-level tag annotations on the dataset, in an effort to promote a stronger geographical diversity in object recognition datasets, many of which are Western-centric.  

The images were kindly annotated by the trainees of the Roia Foundation in Syria.

This object detection dataset is dedicated to the public domain by Humans in the Loop under CC0 1.0 license

cup detection screenshot

Dataset size

The dataset includes 27519 images grouped into 138 folders. 


The images are sorted in 3 types:
1. Abstract – no annotations (18 classes)
2. Places – image-level tags (25 classes)
3. Objects – bounding box annotations (95 classes)

The total number of annotated instances is 32099.

Access the dataset by filling in the form below