Learning Under Low Illumination: A Dataset and Algorithm for Traffic Sign Recognition
Researchers have introduced INTSD, a new large-scale dataset designed to improve traffic sign recognition in low-light conditions, particularly at night. The dataset, collected in India, features diverse nighttime scenarios including headlight glare and motion blur, and covers 41 traffic signboard classes. Alongside the dataset, the team developed LENS-Net, a baseline algorithm that uses adaptive illumination-aware detection and multimodal semantic reasoning to enhance nighttime sign classification accuracy. AI
IMPACT Enhances computer vision capabilities for autonomous systems operating in challenging nighttime conditions.