Target recognition classification covers a broad spectrum of augmented intelligence tools that can quickly identify individual and objects inside images, live video, or a viewfinder. These tools deliver a diverse set of features that can readily classify elements of different sizes, even as small as an item in an individual’s hands. It can also identify shapes, colors, and certain motions. Intellisense has a long history of developing these target recognition and classification systems so that government agencies and armed forces can more effectively complete their work.
When the U.S. military sought an improved automatic target classifier that could identify weaponry, body armor, and still fit onto small-caliber firearms, Intellisense applied its skills in multi-sensor fusion to create the Small Arms Neural Network Automatic Target Classification system, or SANATA for short. Using state-of-the-art artificial intelligence algorithms embedded in a GPU-based edge processor, SANATA can acquire targets, identify weapons or body armor, and display appropriate aimpoints, which greatly improve soldiers’ situational awareness and lethality. This capability can increase soldiers’ responsiveness and accuracy in combat scenarios.
The SANATA system fuses a visible camera, infrared imaging, and an improved deep-learning convolutional neural network (CNN) to identify body armor and differentiate weapons from other objects in a person’s hands. These algorithms were trained to identify these points with data collected under a variety of target ranges and conditions. The long-wave infrared imaging sensor is also capable of working in day, night, or low-light conditions.
Here is just a brief demonstration of SANATA’s capabilities. The video shows two figures in low-light conditions. One is an adversary, and the other is a civilian. The system correctly identifies the enemy combatant’s body armor and rifle, and it can distinguish the weapon from the broom in the civilian’s hands.
In recent testing, SANATA achieved high accuracy and a low false alarm rate in both day and night conditions. Intellisense plans to use funding from the Army Rapid Capabilities and Critical Technologies Office to complete development and conduct qualification testing of the SANATA prototype on an assault rifle. The final prototype will integrate an automatic target recognition algorithm that can detect vehicles and other ground hazards in addition to weapons and body armor. This testing will mature the SANATA system to a Technology Readiness Level 7 or higher.
This development epitomizes Intellisense’s work in automatic target recognition and augmented intelligence. While the deep machine learning algorithms on this rifle-mounted solution help identify enemy combatants and their weak points, a person is still in control of the trigger. This innovation provides guidance to the operator, but the judgment and outcomes are still in human hands. Intellisense strives to continue these multi-sensor fusion developments to keep users informed of nearby dangers and hopefully keep soldiers and civilians out of harm’s way.