Brazilian Urban Mobility Image Dataset

The Brazilian Urban Mobility Image Dataset comprises 1000 JPEG images of Brazilian urban roadways. The image corpus encompasses scenes from four geographically distinct urban centers: Manaus (State of Amazonas), Barueri (State of São Paulo), Chapada (State of Mato Grosso), and Porto Velho (State of Rondônia). This dataset incorporates a total of 6932 annotations, categorized into four classes of objects prevalent in Brazilian urban traffic: passenger cars, motorcycles, buses, and a generalized "vehicle" class (including trucks, vans, ambulances, and similar). Annotations were performed using the Labelbox platform through the application of bounding boxes. The collected dataset was partitioned into three subsets: a training set comprising 60% of the data, a validation set containing 10%, and a test set accounting for the remaining 30%. Object annotations are provided in text files according to the YOLOv11 format.