Military Vehicle Object Detection Based on Feature Representation and Refined Localization Using Inception Recurrent Convolutional Neural Network

Advanced reconnaissance technologies for military vehicle components form basis for weapons and equipment reconnaissance and tracking operations in complex environments, which is crucial to allow intelligent warfare using information. Addressing limitations of conventional detection algorithms in this domain, a novel approach termed as Military Vehicle Object Detection Method based on Inception Recurrent Convolutional Neural Network (MVODM-IRCNN) and Hierarchical Feature Representation is proposed for detecting military vehicle object. first is a comprehensive data set specially designed for military vehicles. Then it is two system improvements: hierarchical feature representation and state-of-the-art localization, using Inception Recurrent Convolutional Neural Network (IRCNN), aimed at improving detection performance. Hierarchical feature representation techniques make it easier to find good feature representation levels for different object shapes, improving accuracy of these techniques boxes. combination of materials provides a significant gain in detector performance. Finally, an exploratory study conducted on own dataset shows that MVODM-IRCNN has exceptional recognition capabilities, which proves its usefulness in military vehicle recognition applications.