The demand for precise automated weapon detection systems has grown significantly in recent years due to heightened security needs and the requirement for rapid threat response protocols. This study proposes an improved weapon detection system by integrating one of the classifier algorithms—K- Nearest Neighbors (KNN), Support Vector Machine (SVM), or Naive Bayes (NB)—with the ResNet50 model for feature extraction. The purpose of this study is to detect and classify two types of weapons—guns and knives—and then evaluate the system by comparing the performance of machine learning algorithms with ResNet50. The ResNet50 deep convolutional neural network, pre-trained on the ImageNet dataset, serves as a tool for extracting complex features from input images in the proposed methodology. The KNN, SVM, and NB algorithms use the extracted features to perform classification tasks. The proposed system is evaluated using the COCO_Guns_Knives dataset, which contains images of guns and knives. The research analysis uses accuracy, precision, recall, and F1-score as evaluation metrics. The highest accuracy of 0.9786 was achieved by combining ResNet50 with the SVM classifier. The research findings demonstrate that combining ResNet50 for feature extraction with SVM for classification provides an effective solution for weapon detection systems. The proposed system can be integrated into security applications to enhance both safety measures and threat response times.