Image datasets for automated bus fault detection

Automated Bus Fault Detection Dataset was created due to the need to study the common visual defects that are typically noticed on buses that run in Bangladesh. The photographs were taken during the time of 40 days, between July and August 2025, in Dhaka City, and were captured mostly with personal smartphones of digital nature. The gadgets that we will use in the collection of our data will include Motorola Edge 20, Xiaomi Note 13 Pro Plus, Redmi Note 12 Pro Max, Redmi Note 10, Samsung A22, and Redmi 12.

All the faces and vehicle license plates seen in the pictures were blurred to avoid revealing confidential information to guarantee privacy and data security. The dataset continued to center on the visible faults of buses and had a curated dataset of around 973 images. These images have six different types of bus conditions that include Wornout_tire, Wornout_body, Broken_glass, Broken_mirror, Emitting_smoke, and Standard_bus. The Standard_bus category was needed in order to compare the defective and the non-defective states.

The image formats are all JPG with the color space being RGB and have different resolutions, with a file size of 150 KB to 300 KB. The given dataset can serve as a good background to the design and testing of the algorithms designed to categorize the visible highway faults based on their evaluation of the deviations in the appearance of the usual standard buses.