🌟 Editor's Note
Welcome to another exciting week in the Vision AI ecosystem! We've got a packed newsletter full of insights, events, and inspiring stories from the heart of innovation.
Top Stories:
Misti AI has raised £250K in pre-seed funding led by Fuel Ventures to transform existing industrial camera networks into real-time AI observability systems. Its edge AI platform converts live video feeds into operational insights, risk alerts, and contextual intelligence for industries like mining, energy, and logistics—bringing “physical AI” to harsh, low-connectivity environments. [link]
Culver’s has partnered with Berry AI to deploy Vision AI across 1,000+ restaurants nationwide. The platform delivers real-time operational visibility into drive-thru performance, throughput, and service execution. The rollout reflects a broader QSR shift from isolated AI pilots to AI-powered operational infrastructure focused on speed, consistency, and guest experience. [link]
The finished vehicle logistics industry is shifting toward real-time, AI-driven operations as supply chain disruptions expose the limits of reactive workflows. At FVL NA 2026, leaders from Overhaul, UVeye, and General Motors — including Nathan Lawson, Yaron Saghiv, and Brian Watanabe — highlighted how AI inspection systems, telematics, and connected vehicle data are improving visibility, damage detection, and yard operations across automotive logistics. (link)
Nota AI won the Grand Prize and Track C at NVIDIA’s first-ever Nemotron Hackathon in Korea, competing against 20 teams including Naver. The team — Hancheol Park, Geonmin Kim, and Geonho Lee — used Nemotron 3 Super 120B to build synthetic data pipelines for MoE quantization, showcasing a data-centric AI optimization approach over traditional algorithm-focused methods. (link)
Startup Spotlight
Berry AI is a Taiwan-based computer vision startup building AI-powered operational intelligence systems for quick-service restaurants (QSRs). Founded in 2019, the company uses camera-based AI, deep learning, and real-time video analytics to optimize drive-thru operations, staffing, throughput, and customer experience
Its flagship products include AI drive-thru timers and dining-room analytics platforms that identify bottlenecks, measure service speed, and improve labor efficiency.
Berry AI is backed by Flytech Technology, an industrial computing company with large-scale deployment expertise. The startup has deployed hundreds of AI-enabled cameras across major global QSR brands, including partnerships with Culver’s and Zaxbys.
Its broader vision is to automate visual operational tasks inside restaurants using AI-powered camera infrastructure integrated with POS and workflow systems.
Paper to Factory
Deep learning models for robust facial liveness detection
This research addresses vulnerabilities in facial recognition systems caused by advanced spoofing attacks, including deepfakes and AI-generated manipulations. The study proposes novel deep learning-based liveness detection models that combine texture analysis with reflective property modeling to differentiate genuine human presence from fraudulent replicas.
Extensive testing across five diverse datasets, covering multiple attack vectors and environmental conditions, demonstrated significant improvements over existing anti-spoofing methods.
The best-performing model, AttackNet V2.2, achieved 99.9% average accuracy when trained on combined datasets while maintaining computational efficiency.
The research also provides insights into impostor attack behaviors, strengthening biometric authentication, identity fraud prevention, and secure access systems across cybersecurity applications. [link]
🏆 Community Spotlight:
Realsense’ latest video with Eric Danziger, CEO of Invisible AI explores the impact of computer vision in factory floors.
The recent RTN Town Hall focuses on the computer vision revolution in the restaurant industry turning video data into actionable intelligence.
A recent talk from University of Stirling focuses on Generative AI for medical computer vision, especially the emergence of synthetic data.
Till next time,
