An Image Dataset for Detection of Tangled Overhead Wires in Urban Environments of Bangladesh

Published: 22 September 2026| Version 1 | DOI: 10.17632/vc938rrf83.1
Contributors:
,
,
,

Description

Rapid urbanization in developing regions has led to the widespread deployment of overhead wiring systems for electricity distribution, telecommunications, and internet services. However, due to unplanned expansions and lack of proper infrastructure management, these wires often become tangled and disorganized. This dataset is a binary image classification dataset containing 3,066 images across two categories: tangled wires and not_tangled wires. The dataset consists of 511 original images, including 402 tangled and 109 not_tangled images. The images were captured using mobile phone cameras at multiple locations across Dhaka, Bangladesh, under different lighting conditions, orientations, and backgrounds. The dataset is intended to support the development and evaluation of machine learning and computer vision models for automated detection of tangled overhead wires. More details and benchmarking performance can be found in our research paper (link is provided below).

Files

Steps to reproduce

Please cite our research paper (link is given below) if you use this dataset.

Institutions

Categories

Image Processing, Urban Planning, Machine Learning, Image Classification, Deep Learning

Licence