Assaults using weapons like pistol and knife are common in today’s world. Many weapon detection models are available for detecting weapons but there is only limited attention given to visual weapon detection during night or at low lighting conditions. This work suggests a weapon detection system which works in dark and low light conditions to detect weapon. Datasets consisting of pistol and knives are used for training the YOLOv8 detection model. In dark regions the model works by first enhancing the image using an image enhancement algorithm which brightens the surrounding and then send to the detection model for weapon detection. The model gives a mAP of 93.6 for normal lighting conditions and mAP of 81.1 for enhanced dark images using the proposed model. YOLOv8 model is compared with Faster RCNN and have an increase of 5.9% in mAP.