An Image Dataset for Detecting Unauthorized Vehicle Parking in Urban Roads of Bangladesh

Published: 24 September 2026| Version 1 | DOI: 10.17632/sv8hbtbmjn.1
Contributors:
Nuran Nahadi Islam,
,
,
, Ashraful Alam Ashraful Alam,

Description

This dataset contains 443 original, hand-collected and manually annotated street-level photographs of narrow two-lane urban roads in Bangladesh. It was created to support research on automated detection of unauthorized vehicle parking — a persistent cause of traffic congestion in densely populated cities like Dhaka. Each image is labeled as either Positive (a four-wheeled vehicle parked on a narrow road so that another vehicle cannot pass — unauthorized/blocking) or Negative (authorized parking, non-blocking scenarios, empty roads, moving vehicles, two/three-wheelers, or hard-negative scenes that resemble violations but are not). The Positive class contains 224 images and the Negative class contains 219 images.

Files

Steps to reproduce

Please cite our research paper (link is provided below) if you use the dataset.

Institutions

Categories

Computer Science, Computer Vision, Urban Planning, Transportation Engineering, Image Classification, Deep Learning

Licence