Sophisticated Weapon Detection Algorithms: A Quintuple Class Approach Using CNN-SVM

To obtain a metric precision of 88.33% and a metric recall of 87.13%, the Fl-Score being 87.72%. The data shows three classes separated as Class 1, which occupies 20% of the dataset (1903 cases) out of 9462 cases in the set. For Class 1, we produced an effect of 95%. As a Class 2, its precision reached 88.71 % with a recall value of 93.17% and an F1 score of 90.88%, where 1830 cases with 19% coverage of the dataset were considered, and overall accuracy achieved 96%. The performance is marked by an accuracy rate of 89.98%, a recall rate of 89.1 %, and an Fl-Score of 89.54. This group saw the count of 1935, which is a fifth of all data set. Briefly, the Class 3 precision is 96%. The class 4 precision level stands at 90.03%, the recall level is 92.61 %, and the Fl-Score is 91.3%, with 1881 cases corresponding to 20% of the data set and the eventual model accuracy at 97%. Class 5 stands for its specificity regarding the high marks achieved. The accuracy is 94.4%, the recall is 89.57%, and the Fl-Score is 91.92%. Moreover, not only is it the most common one with 1975 occurrences that constitute 21 % percent of the file, but it also has the most extensive support out of all the words. Lastly, it meets a record accuracy of 97%. The paper, which leans on macro, micro, and weighted averages, depicts the general output of the model. The macro means accuracy, recall, and Fl-Score are approximately 0.902940395, 0.903121043, and 0.902721029. This said, the model is uniformly triumphant in each category.