4/26 ARROW RELEASE OF CIGNAL AI NOW AVAILABLE FOR GOVERNMENT USE
REQUEST ACCESS
April 06, 2026. - Cignal has been selected to present at SPIE ADIX X. Cignal's research on Vision Language Models is entitled: Let's be reasonable: why compositional VLMs are the future of ATR.
REEDSVILLE, PA, April 06, 2026 - Cignal has been selected to give a presentation at SPIE ADIX entitled: "Let's be reasonable: why compositional VLMs are the future of ATR." This talk will explore how Vision-language models (VLMs) offer the X-ray security imaging domain considerable benefit in the form of compositional thinking. Not limited to a single pass, VLMs have the natural capacity for iterative analysis, conditioning on input as well as previous actions and observations to arrive at solutions for complex tasks. Their large vocabularies and demonstrated language reasoning abilities not only enable intelligent behavior, but also ground their vision modules and avoid the brittleness of vision-only models.
Read more on the SPIE ADIX conference page.
