The applications for AI are increasing not only in the civilian sector, but also in the military domain. Object detection, classification and tracking are some of the relevant applications for defence purposes.
Dataset for Military Objects
Training an AI model to detect military objects requires an extensive dataset. The dataset must contain a large number of labelled images. As there often is only limited open source data available, we developed a synthetic training data generation pipeline to create the necessary datasets.
Whenever possible, we complement the synthetically generated training data with as many real and manually labelled images as available, to increase the accuracy of the trained AI model.
Sample Synthetic Images
AMDC Military Object Detection is based on state-of-the-art Object Detection Model Architectures, such sas YOLO, SSD or FasterRCNN, which we carefully modify and customly train to satisfy the challenging requirements of military applications. Our models are specifically optimized to provide a very high accuracy and, at the same time, allow for sufficient refresh rates on current embedded systems, such as NVidia Jetson Modules or the latest FPGAs.
Starting in 2018 AMDC develops and trains object detectors that can classify tanks along with other objects such as people, trucks, cars, motorcycles, and others. The video on the right shows the “Proof of Concept”, detecting tanks in different publicly available footage.