“Enhanced Object Detection in Low-Light Conditions Using Modified Yolo Architecture for Night Surveillance”

This has made night surveillance an irreplaceable part of the security systems, yet object identification under low-light situations is a significant problem owing to noise, low contrast, and blurred objects. Standard deep learning models and traditional computer vision models are not generally able to provide believable accuracy in such circumstances. This paper suggests a modified YOLO (You Only Look Once) architecture, which can be used to detect objects in low-light surveillance systems. The model adds image preprocessing, attention block, and light convolutional blocks to enhance the precision of detection and simultaneously run in real-time. The experimental findings show that the suggested framework is much more effective than the used baseline YOLO models regarding mean Average Precision (mAP) and detection speed (FPS), as well as low-illumination resistance