Ambarella and Ultralytics Collaborate to Bring Ultralytics YOLO Models to CVflow-Powered Edge Devices

Ambarella and Ultralytics plan to streamline deploying YOLO vision models on low-power CVflow edge AI hardware for on-device inference.

Rhea-AI Summary

Ambarella (AMBA)/b) and Ultralytics announced a collaboration to deploy Ultralytics YOLO computer vision models on Ambarella’s CVflow-powered edge AI devices.The partnership combines Ultralytics’ real-time vision models with Ambarella’s low-power AI processing to support on-device detection, segmentation, tracking, classification, pose estimation, and oriented bounding box detection. Running these workloads locally is intended to lower latency, reduce network bandwidth needs, and lessen reliance on cloud processing, which can help systems operate more reliably when connectivity is limited.

The integration targets developers and product teams building edge applications for smart cameras, robotics, automotive, and industrial systems, using Ambarella’s Cooper Developer Platform and DevZone, which provide optimized models, blueprints, kits, and unified software across its edge AI SoC portfolio.

Market Context

AMBA's pre-headline change was -5.72%, while all five listed peers were also lower at publication; t...

Analysis

AMBA's pre-headline change was -5.72%, while all five listed peers were also lower at publication; the collaboration therefore arrived amid aligned weakness across the tracked peer group.

Key Terms

object detection, instance segmentation, pose estimation, oriented bounding box detection

4 terms

object detection technical

"Ultralytics YOLO models support object detection, instance segmentation"

Object detection is software that finds and identifies specific items within images or video and marks where they are, like a smart camera that draws boxes around cars, medical signs or products on a shelf. For investors, it signals how companies can automate inspection, monitoring or compliance tasks, reduce labor and error, and create new product capabilities that may drive revenue, cost savings and competitive advantage.

instance segmentation technical

"YOLO models support object detection, instance segmentation, image classification"

A computer vision method that detects and outlines each individual object of interest in an image, giving a separate pixel-accurate mask for every instance rather than just labeling object types. Think of it as not only saying “there are three cars” but drawing the exact shape of each car; investors care because it is a concrete technical capability underlying products, performance claims, and market applications for AI-driven imaging, automation, and compliance tools.

pose estimation technical

"image classification, pose estimation, and oriented bounding box detection"

Pose estimation is a computer vision technique that identifies the position and orientation of a person or object in an image or video, often by locating key points such as joints on a body or corners on an object. For investors, it matters because it measures a company’s ability to turn camera data into actionable products—for example, in robotics, augmented reality, surveillance, or healthcare—so it can signal technical capability, product applicability, and market opportunity.

oriented bounding box detection technical

"pose estimation, and oriented bounding box detection within a single model family"

Detection that locates objects in images by drawing rectangular boxes that can rotate to match an object's angle, rather than only using upright, axis-aligned rectangles. It matters to investors because rotated boxes give more accurate size, position and orientation information for use cases like aerial or satellite imagery, automated inspection, and document analysis—think of fitting a tilted picture frame around a slanted photo so measurements and counts are more precise.

AI-generated analysis. How Rhea-AI works. Not financial advice.

See more from StockTitan in Google Search and AI answers. Adds StockTitan as a preferred source · opens Google

Add on Google Not now

The partnership is designed to give developers and product teams a more efficient path from vision model development to deployment on low-power edge AI silicon

SANTA CLARA, Calif. and LONDON, Sept. 15, 2026 (GLOBE NEWSWIRE) -- Ambarella, Inc. (NASDAQ: AMBA), an edge AI semiconductor company, and Ultralytics, a vision AI company known for the open-source Ultralytics YOLO model family, today announced a collaboration to support the deployment of Ultralytics YOLO models on Ambarella CVflow®-powered edge devices.

The collaboration brings together Ultralytics' real-time computer vision models and Ambarella's low-power AI processing technology. The two teams are working to give developers a familiar starting point on silicon built for the power and thermal limitations under which edge products operate.

When detection, segmentation, and tracking run locally on the device, systems can respond with lower latency, place less demand on network bandwidth, and reduce their dependence on cloud processing. For products that operate in the physical world, those characteristics shape how quickly a machine responds to what it sees, and how dependably it responds when connectivity is limited.

The integration is designed to help developers and product teams build and deploy vision applications more efficiently across areas such as smart cameras, robotics, automotive, and industrial systems.

"By bringing Ultralytics YOLO to Ambarella's CVflow platform, we're making it easier for developers and product teams to build efficient, real-time vision applications directly at the edge," said Glenn Jocher, Founder and CEO of Ultralytics. "This collaboration combines Ultralytics' accessible computer vision models with Ambarella's low-power AI processing, helping move vision AI from development to deployment more efficiently."

On the Ambarella side, the work centers on the Cooper™ Developer Platform and the Ambarella Developer Zone (“DevZone”), which the company launched at CES 2026. The DevZone brings optimized models, agentic blueprints, developer kits, documentation, and technical resources into a single accessible environment. A common CVflow architecture and a shared software environment serve Ambarella’s entire edge AI SoC portfolio, so perception software and development workflows can carry forward as a product line scales up in performance or moves to a lower cost point.

"The edge AI markets we serve are evolving quickly, and our partner ecosystem needs a platform that meets developers in the frameworks and model families they already use," said Fermi Wang, President and CEO of Ambarella. "Ultralytics YOLO is among the most widely adopted model families in computer vision, and supporting it on our CVflow platform broadens the set of teams that can build on Ambarella silicon."

Ultralytics YOLO models support object detection, instance segmentation, image classification, pose estimation, and oriented bounding box detection within a single model family, with a range of model sizes that let teams balance accuracy against the compute and memory budget a given product can carry.

About Ambarella

With an installed base of more than 50 million AI SoC units, Ambarella's products are utilized in a wide variety of physical edge AI applications, spanning edge endpoint and edge infrastructure use cases including physical security, vehicle safety, telematics, autonomy, portable video, aerial drones, and other emerging robotic applications. Building on this footprint, Ambarella offers a full-stack edge AI platform, from highly optimized silicon and programmable software to AI agentic frameworks that coordinate perception, decision-making, and control across devices. Ambarella's low-power systems-on-chip (SoCs) integrate proprietary and highly efficient perception and deep learning neural network AI accelerators, enabling electronic systems to become more productive with partial or complete levels of machine autonomy. For more information, please visit www.ambarella.com.

About Ultralytics

Ultralytics is the leading force in vision AI, best known for its Ultralytics YOLO (You Only Look Once) models and Ultralytics Platform, the ultimate end-to-end computer vision platform where users can annotate, train, and deploy vision AI solutions in one place. With over 140k GitHub stars, 330 million PiP downloads, and 3.6 billion model usages, Ultralytics YOLO has become the most widely recognized object detection model globally. Ultralytics empowers users with easy-to-use, cutting-edge AI technology. Our mission is to simplify and democratize the use of AI, making it accessible and impactful across industries ranging from manufacturing to healthcare and more.

Media Contact

Jonathan Miller

jmiller@ambarella.com

FAQ

AI-generated questions and answers. How Rhea-AI works. Not financial advice.

Which Ambarella tools and resources are involved in the Ultralytics YOLO integration?

The work focuses on Ambarella’s Cooper Developer Platform and the Ambarella Developer Zone (DevZone), which bring together optimized models, agentic blueprints, developer kits, documentation, and technical resources in a single environment. A common CVflow architecture and shared software environment span Ambarella’s edge AI SoC portfolio so perception software and workflows can carry forward as products scale performance or move to lower cost points.

What types of vision tasks do Ultralytics YOLO models support on Ambarella CVflow devices?

Ultralytics YOLO models on Ambarella CVflow devices support object detection, instance segmentation, image classification, pose estimation, and oriented bounding box detection within a single model family, with multiple model sizes to balance accuracy against compute and memory budgets.