CAMOUFLAGED MILITARY PERSONNEL DETECTION USING SSIM-BASED FASTER R-CNN MODEL

In modern defence operations, accurate and timely detection of hidden targets is crucial for personnel safety and mission success.Camouflaged military personnel are currently difficult to detect using traditional surveillance systems, as their appearance often blends with complex natural environments.To highlight this challenge, an improved deep learning model named the Structural Similarity Index Measure (SSIM)based Faster Region-Based Convolutional Neural Network (Faster R-CNN) is proposed in this study.Here, the SSIM loss guides the Faster R-CNN model to focus more on subtle differences in structure and texture.This process helps in detecting hidden targets (camouflaged military personnel) more accurately.The performance evaluation on a real-time dataset confirmed that the proposed SSIM-based Faster R-CNN model achieved better results compared to traditional detection models in terms of mean Average Precision (mAP) and Intersection over Union (IoU).The experimental results obtained from three different environments prove that the proposed SSIM-based Faster R-CNN model is suitable for real-time use in modern defence and surveillance systems.