A Multi-Branch Network for Automated VNIIRS Assessment of Motion Imagery

Due to the lack of consistency in existing automated methods for assigning VNIIRS levels to motion imagery, and the overwhelming human resources required to manually assign levels, a new method of automated/semi-automated VNIIRS assessment is needed. In recent years, advancements in deep learning have provided solutions to previously intractable computer vision problems. In many cases, automated deep learning algorithms have been able to meet and in some cases surpass human analysis performance for well-defined object classification tasks. Deep learning algorithms are also extremely efficient and offer scalability for solving challenging large scale problems. For these reasons, a deep learning solution should be considered. Toyon proposes a automating the assignment of VNIIRS labels to motion imagery by using a combining motion detection, object and event classification that will provide information to a final classification a network that decides the VNIIRS level of a motion imagery clip. These classification techniques leverage a novel multi-branch convolutional neural network and state-of-the-art long-short term memory networks. The system will enable intermediate textual outputs to be displayed to analysts to improve their understanding of the rating and give additional order of battle context for the processed motion imagery.