Related Concept Videos
Light Acquisition
Difference from Background: Limit of Detection
The LOD indicates the presence or absence...
Detection of Black Holes
Their closest cousins are neutron stars, which are composed almost entirely of neutrons packed against each other, making them extremely dense. A neutron star has the same mass as the Sun but its diameter is only a few kilometers. Therefore, the escape velocity from their surface is close to the speed of light.
Not until the 1960s, when the first neutron...
You might also read
Related Articles
Articles linked to this work by shared authors, journal, and citation graph.
Editorial: Harnessing visual computing to revolutionize manufacturing efficiency and innovation.
PIO, A Large-Scale Dataset for Broiler Chicken Detection under Real Poultry Farming Conditions.
Artificial intelligence as a tool for bionomic transects: the case of Isidella elongata (Esper, 1788) forests in the Western Mediterranean.
Related Experiment Video
Updated: Sep 24, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
UrOAC: Urban objects in any-light conditions.
Francisco Gomez-Donoso1, Marcos Moreno-Martinez1, Miguel Cazorla1
1University Institute for Computer Research, University of Alicante. PO Box 99, Alicante 03080, Spain.
This study introduces UrOAC, a new dataset for urban object detection under diverse lighting. It enables AI systems to better detect traffic lights and crosswalks in low-light and nighttime conditions.
Area of Science:
- Computer Vision
- Artificial Intelligence
- Robotics
Background:
- Existing urban object detection methods fail in low-light conditions.
- Current datasets lack diversity in lighting, limiting AI performance at night.
- Visually impaired assistance and autonomous vehicles require robust nighttime object recognition.
Purpose of the Study:
- Introduce UrOAC, a novel dataset for urban object detection.
- Capture urban scenes across various lighting, from daylight to nighttime.
- Provide annotated data for pedestrian crosswalks, and red/green traffic lights.
Main Methods:
- Collected images of urban objects under diverse lighting conditions.
- Annotated objects with categories and bounding-boxes.
- Developed the UrOAC dataset for training and evaluating AI models.
Main Results:
- UrOAC dataset features urban objects in varied lighting, including nighttime.
- Annotations include object category and precise bounding-boxes.
- The dataset addresses the gap in low-light urban scene understanding.
Conclusions:
- UrOAC dataset can significantly improve nighttime and low-light performance of vision-based systems.
- Enhances object detection for visually impaired assistance devices.
- Supports the development of safer self-driving and intelligent vehicles.

