L-VAD (Low-Light Vehicle & Annotation Dataset)

Published: 14 January 2026| Version 2 | DOI: 10.17632/h6p2w53my5.2
Contributors:
ade kurniawan,
,
,
,
,

Description

Overview This dataset is a specialized collection of annotated images designed for Nighttime Vehicle Detection. While most standard datasets focus on daylight conditions, this dataset addresses the specific challenges of low-light environments, such as glare from headlights, motion blur, and low contrast. It provides high-quality images of three primary vehicle classes: Cars, Motorcycles, and Trucks. Every image has been meticulously annotated with bounding boxes, making it ready for supervised learning tasks in Computer Vision. Key Features • Environment: Exclusively nighttime and low-light conditions (urban streets, highways, and residential areas). • Object Classes: 1. Cars (Sedans, SUVs, Vans) 2. Motorcycles (Scooters, Sportbikes) 3. Trucks (Freight trucks, Delivery trucks) • Annotation Format: Provided in YOLO format. • Diversity: Captured under various night conditions, including street-lit roads and total darkness with only vehicle headlights as light sources.

Files

Categories

Artificial Intelligence, Computer Vision, Transportation Engineering, Pattern Recognition

Licence