A Survey on Enhanced Weapon Detection System using Deep Learning

Considering the increasing prevalence of criminal activities, this study introduces a novel deep learning model designed for accurate weapon detection. The model, based on the VGGNet architecture and implemented on TensorFlow, classifies seven weapon types: assault rifles, bazookas, grenades, hunting rifles, knives, pistols, and revolvers. Trained on a custom dataset, the model achieves an impressive 98.40% accuracy, surpassing VGG-16, ResNet-50, and ResNet-101. The proposed system addresses challenges like self-occlusion and complex backgrounds, offering practical applications for enhancing security.