SOF Week 2026: US military tests AI algorithm to support missions in low-light scenarios
Image showing the difference between standard night vision and Deepnight Night Vision. (Photo: Deepnight)
The US Army and USAF are evaluating an AI-enabled imaging capability from Deepnight designed to enhance low-light and no-light operations across multiple platforms and environments.
The US Army and US Air Force (USAF) have conducted tests with an artificial intelligence (AI) algorithm designed to facilitate deployments in low- and no-light conditions.
Supplied by California-based company Deepnight, the software solution has been developed to increase photosensitivity and performance for night vision systems.
The trials and demonstrations have been carried out since last year, with the most recent taking place in April at USAF Barksdale Air Force Base in Louisiana and the Barry M Goldwater Range in Arizona.
Related Articles
How commercial tech is reshaping military requirements
L3Harris awarded $263 million contract for night vision goggles
Speaking to Shephard, the company’s CEO, Lucas Young, noted the solution has been demonstrated to the US military across diverse platforms and environments.
“We have basically taken this to almost every possible domain. We have deployed this underwater, aerially on a drone, on people’s heads,” he said.
Over the coming months, the US Army will conduct additional experimentation with the solution under the Soldier Borne Mission Command (SBMC) programme, previously known as the Integrated Visual Augmentation System (IVAS).
Designed to recover obscured data and improve the performance of conventional systems in real time, the capability can be integrated into thermal and infrared cameras, handheld devices, robots and autonomous vehicles.
It can also be attached to soldiers’ helmets and fitted into in-service platforms to process imagery in dark scenarios. Young claimed the solution processes “very limited signal and renders what that would look like if we were capturing in the daytime”.
“The core piece of the [Deepnight’s] technology is the AI algorithm. So, we have reformulated the night vision problem primarily as a computational problem,” Young claimed.
The capability works at speeds from 60 to 90 frames per second while requiring limited edge compute. With this technology, the company intends to fill a gap in the defence market by reducing the costs associated with the acquisition and maintenance of night vision capabilities and unlocking consumer electronics supply chains for mass production.
From Young’s perspective, it represents a “major technical milestone” by enabling the capture of images “in the extremely low light level with minimal latency”.
“PVS-9 and PVS-14 [military designations for night vision devices] are extremely expensive and hard to produce. The entire idea of Deepnight is to basically commoditise it and turn it into something very scalable and then bring night vision to basically every platform in the military,” he added.
To advance the implementation of this technology, the vendor has been working with other suppliers, such as Eolian, Sionyx, SRI International, Devcom, Circle Optics and Picogrid.
More from Defence Notes
-
![Industrial capacity under scrutiny as US approves further $8.6 billion Middle East arms sale]()
Industrial capacity under scrutiny as US approves further $8.6 billion Middle East arms sale
The fast-tracked emergency approvals come as the conflict in the Middle East stretches out into its third month, after Iranian attacks depleted US allies’ missile stockpiles and testing air defence systems.
-
![Intelligence innovation: From data overload to decision advantage (Podcast)]()
Intelligence innovation: From data overload to decision advantage (Podcast)
As militaries face an overwhelming flow of data, the challenge is shifting from collection to delivering fast, actionable insights that drive decision-making. Advances in AI and data integration are helping armed forces move beyond siloed systems to generate real-time intelligence across domains and allies.
-
![SAHA 2026 to Convene the Global Defence Ecosystem]()
SAHA 2026 to Convene the Global Defence Ecosystem
SAHA 2026 brings global defence and aerospace leaders to Istanbul for partnerships, launches, panels and high-value meetings.
-
![Teledyne FLIR adds GPS-denied 3D-mapping capabilities to its CBRN uncrewed platforms]()
Teledyne FLIR adds GPS-denied 3D-mapping capabilities to its CBRN uncrewed platforms
In a partnership with Emesent, Teledyne FLIR will equip its autonomous air, ground and detection systems with the Hovermap LiDAR payload in a move that highlights a broader market shift towards modular architectures, shared payloads and interoperability across platforms.
-
![US seeks 32% boost for missile defence budget with $23 billion earmarked for interceptors]()
US seeks 32% boost for missile defence budget with $23 billion earmarked for interceptors
The Pentagon’s proposed budget for the next fiscal year includes an impressive increase in the procurement of interceptors, with the number of the US Army’s PAC-3 MSE rounds expanding by 683%, the US Navy’s Standard Missile by 365% and the MDA’s SM-3 IIA by more than 1,000%.
-
![US Army partners with Global Military Products to surge munitions production]()
US Army partners with Global Military Products to surge munitions production
Global Military Products was selected by the US Army to operate the Quad Cities Cartridge Case Facility and ramp up the production of various calibre shell cases.
