News
From pixels to perception: Teledyne FLIR redefines thermal imaging at DSEI
“It’s about maturity and immediacy:
proven cores, plus software that
adds real operational value.”
Interview with Jared Faraudo, Vice-President of Product Management
Boson, Prism and the new integration imperative
DSEI, ExCeL London
There was a telling moment on Teledyne FLIR’s stand at DSEI. Jared Faraudo, Vice-President of Product Management, leaned forward and tapped an image on the demo screen: one frame showed the raw thermal output from a cooled mid-wave sensor; alongside it, the same scene after Teledyne’s Prism processing and edge AI had been applied. The change was stark. Edges sharpened, background clutter dropped away, and moving contacts stood out as clean, trackable objects. “That’s the point,” Faraudo said. “You don’t just hand somebody a sensor any more. You hand them a capability.”
Boson: the integrator’s staple
The Boson family, Teledyne’s long-wave infrared modules, has become a mainstay for system integrators seeking compact, low-power thermal cores. Small, robust and supported by a wide ecosystem, Boson’s Vanadium-Oxide microbolometer variants (320 × 256 and 640 × 512 formats) are designed for straightforward integration. Radiometric options and a broad lens suite broaden the applications. “Boson is deliberately integrator-friendly,” Faraudo explained. “It’s SWaP-optimised, it’s supported by an SDK, and it’s designed to reduce time-to-field.”
Prism: shifting the value chain
What has changed this year is not the sensor itself but the way Teledyne is binding hardware and software. Prism, the company’s image signal processing and AI environment, is aimed at removing the long, costly AI development phase that once sat between data collection and operational output. “Customers told us they didn’t want to spend two years and millions building an AI pipeline,” Faraudo said. “They wanted something that worked with the platforms they already have.”
Prism delivers denoising, super-resolution and pre-trained perception models that run on embedded processors. The result is edge inference and object tracking without the overhead of a bespoke engineering programme.
Neutrino and the high-end calculus
Pairing Prism with Teledyne’s Neutrino mid-wave cores, notably the SX8, a 1,280 × 1,024 HOT MWIR device with an 8-micron pitch, produces striking results. Neutrino serves airborne, ISR and gimballed systems where sensitivity, range and continuous zoom optics are critical. Adding computational super-resolution and detection algorithms extends practical recognition and tracking without redesigning the optics. “You can get more useful image out of the same aperture and make better decisions for the platform,” Faraudo noted.
For programme managers, the attraction is clear. A Boson core plus a Prism stack represents a very different proposition from a sensor coupled with a bespoke AI contract. Faraudo presented it as risk reduction: “Buy a tested thermal module, buy a tested software stack, and you collapse months of verification into weeks. That matters when you are working under a schedule or a limited integration window.”
He was quick to temper expectations: “Pre-trained models get you most of the way; for specialised targets you still do the dataset work. Prism is designed to make that dataset, tuning and validation phase far faster.”
Smarter processing, not magic
Technically, Prism’s image processing implements denoising and super-resolution routines that increase perceived detail, while its AI layer provides classification and multi-object tracking at the edge. The net effect is often described as extending the useful range of a thermal sensor, particularly attractive in small UAS, counter-UAS, ground ISR and maritime surveillance where payload, power and heat budgets are tight. “We’re seeing customers trade a more complicated optical build for smarter processing,” Faraudo observed.
Markets and limits
Faraudo acknowledged the dual-track market. Commercial sectors such as traffic monitoring, industrial inspection and site security continue to support Boson and Boson+, while the Neutrino line is aligned with defence and advanced autonomy programmes demanding MWIR performance. “We design the portfolio to be flexible across civil and defence,” he said, “but we also provide tools so a system integrator can get to field trials sooner.” Yet Prism does not replace optics. “You don’t replace a lens with software,” Faraudo admitted. “You extend what the lens can deliver in realistic, operational scenes.”
Procurement pressures
At DSEI, the offering looked less like a vendor pitch and more like a response to today’s procurement realities: shorter development cycles, constrained budgets and demand for modular, upgradeable architectures. For customers seeking rapid capability insertion, the Boson line, Prism suite and Neutrino cores together present a persuasive case: maturity in the sensor, immediacy at the edge.
Editor’s Notes
Teledyne FLIR’s strategy reflects a wider trend in electro-optics: shifting value from hardware to edge-ready, software-defined capability. Boson remains the integrator’s choice for compact cores in commercial and lower-tier defence roles, while Neutrino serves higher-end ISR and airborne surveillance.
Prism is the differentiator. By bundling image processing and pre-trained AI models, it promises faster integration and reduced programme risk, a direct response to NATO procurement pressures for shorter cycles, lower cost and modular upgrades.
In civil markets Boson and Boson+ paired with Prism remain competitive, but the real growth lies in defence, where governments are seeking scalable architectures that combine proven sensors with deployable AI.
