Development of an Algorithm for Detecting Hidden Targets Using Artificial Intelligence

Target detection is vital in security monitoring, military reconnaissance, remote sensing, and other activities. Accurate detection of camouflaged or hidden targets is critical to effective operations behind the wall, but traditional human observation and simple image processing methods are often inadequate in harsh environments. This work investigates using convolutional neural networks (CNN) to detect hidden targets in complex backgrounds. We aim to overcome the limitations of traditional technologies by leveraging CNNs that revolutionise image recognition and object detection. We build a CNN model with multiple convolutional layers, batch normalisation, ReLU activation, and max pooling, and finally create layers and a softmax layer for classification. Adam Optimiser is used to optimise the existing model. The model was evaluated using performance metrics such as an accuracy value of 0.89067 and an F1 score of 0.94254. This model has high precision and broad prospects in practical applications compared to conventional models. This study demonstrates the feasibility of an intelligence-driven approach for advanced reconnaissance in various battlefield environments. The effectiveness of the developed model has been verified for synthetic as well as real-time Through Wall Imaging (TWI) data and has produced promising results.