AI Computer Vision for the Retail Industry
Accurate, High-Speed,
Product Identification, Powered by AI
Delivering innovation with AI models designed and trained for retail
AI-based product identification that just works. At scale. At speed. In the real world.
Identify hundreds of thousands of products quickly and accurately from any angle, in any lighting, at scale, and despite obstructions and variants.
UltronAI enables you to solve today’s most pressing retail problems by harnessing the power of artificial intelligence with a computer vision platform purpose-built for the retail industry.
Trained to work in highly variable store conditions, our industry-leading product identification delivers confidence and efficiency to any—and every—product-related use case.
The unfulfilled promise of computer vision
It’s no secret that previous AI-based visual product identification solutions have fallen short. Computer vision’s promise has often faltered due to:
1
Real-world conditions
Enormous amounts of data sampling to address issues with occlusion, position, lighting, shadows, and fine-grained details
2
Volume limitations
Inability to scale to meet enterprise needs in an economical way
3
Long training cycles
Training, expanding, and retraining required significant investments of time and money
4
High operating costs
Costly dedicated servers or cloud computing resources required to run AI
Finally
AI product identification you can trust.
Fast Identification
Reduce errors, friction, and customer frustration with instant product identification
Fast Enrollment
Simplify support for large, changing product catalogs with one-shot enrollment
Accurate
Achieve unparalleled accuracy with a retail-trained AI model built for real-world conditions
Scalable
Support hundreds of thousands of products at speed
Easy to Use
Create, refine, and expand the model without needing AI expertise
Easy to Deploy
Speed development and time-to-value with an SDK and API-first architecture that works either on the edge or in the cloud
Validated by a top US retailer
One of the most prominent retailers in the US, with 4,000+ locations across the country, has turned to UltronAI to decrease check out times and protect against losses from theft.
Prior to implementing UltronAI, the company was using an internally developed system that could only recognize 1,000 products. With UltronAI, 250,000 products were onboarded in under an hour—with near 100% identification accuracy. That’s the power of UltronAI.
| Capability | Typical AI Limitations | The UltronAI Platform for Retail | UltronAI Innovations |
| Catalog Enrollment Time | Days to weeks to enroll 1-3k products | 250k products < 45 minutes* | Zero-shot, single catalog image enrollment vs multi-image, multi-view capture |
| Gallery Size (Number of Products Successfully Enrolled) | 1-3k products | 250,000+ products* | Foundational AI models built explicitly to maintain accuracy and speed as product catalogs grow |
| Product Identification Accuracy | Variable lighting and positioning conditions negatively impact accuracy | Over 99%† | AI trained specifically for robustness to occlusion, pose, lighting, and fine-grained similarities |
The UltronAI Platform
UltronAI isn’t a “better scanner”—it’s an embeddable AI engine that enables retail innovators to reimagine what’s possible when it comes to product detection and identification.
Built off 20 years of military-grade artificial intelligence research and more than 50 patents, the AI computer vision models underlying our product detection and identification engine are purpose-built and trained from the ground up for the retail industry.
Designed to lift traditional barriers around speed, scale, and ”ideal” conditions, our platform enables new customer innovations—around high-speed checkout, automation, loss prevention, inventory management, and more—to move from idea to reality.
UltronAI
Identify
UltronAI accurately identifies any packaged product in your catalog
UltronAI
Detect
No need for barcodes, it works with limited angles and partial views
UltronAI
Gallery
Quickly add thousands of products to your catalog within hours
UltronAI
Integrate
UltronAI works alongside your existing software platforms with no conflicts
UltronAI
Mobile
Use the UltronAI mobile app to quickly add new products with enterprise-wide real-time updates
UltronAI
Dashboard
Get real-time reports on accuracy in a customizable dashboard
Your retail vision.
Made possible by AI computer vision.
The technology behind UltronAI
Backed by 50+ patents, UltronAI’s technology is based on more than 20 years of military-grade R&D focused on accurate identification from partial images.
Marios Savvides, PhD, our Chief Technology Officer, boasts a distinguished background in facial identification technology. This pioneering AI work in large scale visual identification has lead him to receive the “Outstanding Contributor in AI” from the U.S. Secretary of the Army, Mr. Ryan McCarthy in 2020. In 2022 he was named “Inventor of the Year” by the Pittsburgh Intellectual Property Law Association for his large and extensive patent portfolio. These awards are in addition to 6 other awards he has received for his research impact. This groundbreaking proprietary IP was adapted and enhanced to form the core of UltronAI’s specialized AI proprietary computer vision models for retail product identification.
While facial identification might seem a far cry from correctly identifying the ketchup that’s hiding behind the mayo, at a computer vision level, both problems rely on taking a partial or distorted picture and attributing it to the right real-world object. Today, UltronAI delivers the most accurate computer vision engines for product identification by building on this foundation.
UltronAI Leadership
Marios Savvides
Founder, Chairman, and CTO
