Deep Learning Algorithms for Multi-Class War Event Classification

Recent years have seen significant progress in using deep learning (DL) for image processing. However, further investigation is required to effectively use DL approaches for multiclass images. These advancements include using computer vision tasks and machine learning methods to address existing issues and provide groundbreaking solutions for the preservation and surveillance of military events. The objective of this research is to categorize images depicting combat incidents. The dataset used in this investigation comprises 1100 images depicting battle incidents. The research utilizes convolutional neural networks (CNNs) to create multiclass classification models. The CNN was optimized via transfer learning and fine-tuning of hyper parameters. The results indicate a significant level of precision, as the validation test revealed an overall highest accuracy of 92.22%, with F1-score values above 89% for the VGG19 model with compare to VGG16 and ResNet50.