UAV_SMID: UAV Sky Monitoring Image Dataset

Published: 8 July 2026| Version 1 | DOI: 10.17632/3k3hjc7rkt.1
Contributors:
Md Zahurul Haque,
,
,
,

Description

This dataset presents a comprehensive collection of 13,928 aerial images containing 16,229 annotated objects, specifically curated for Unmanned Aerial Vehicle (UAV) sky monitoring and airspace security. The dataset features five distinct classes: helicopter, bomb, drone, bird, and aeroplane. A key advantage of this dataset is its highly balanced class distribution, with object counts ranging from 3,162 to 3,440 per class. This balance makes it an ideal benchmark for training robust deep learning and machine learning models for aerial object detection, threat identification, and reducing false positives by distinguishing between aerial threats (drones, bombs) and natural objects (birds).

Files

Institutions

Categories

Computer Vision Technology, Deep Learning Image Reconstruction

Licence