Edge AI and Vision Insights: March 12, 2025

 LETTER FROM THE EDITOR

Dear Colleague,

Vision-language models (VLMs) promise to revolutionize computer vision, but how can you leverage them in real-world applications? If you’re an engineer, developer or engineering manager eager to take advantage of this new technology, join us for an intensive three-hour training session taking place in Santa Clara, California on May 20, the first day of the 2025 Embedded Vision Summit.

This training is designed to introduce the latest techniques in VLMs and their integration with traditional computer vision methods. With a focus on the practical application of these techniques for real-world applications, this course is tailored for professionals looking to expand their skill set in AI-driven computer vision, particularly in systems designed for deployment at the edge.

The training will be led by Satya Mallick, CEO of OpenCV.org and founder of Big Vision LLC, and Jeff Bier, President of BDTI and founder of the Edge AI and Vision Alliance. See here to learn more and register. And while you’re there, don’t forget to register for the Embedded Vision Summit main program, too!

Brian Dipert
Editor-In-Chief, Edge AI and Vision Alliance

AI DEVELOPMENT CASE STUDIES

Ten Commandments for Building a Vision AI Product

Over the past three decades, the convergence of machine learning, big data and enhanced computing power has transformed the field of computer vision from basic image and signal processing to complex perception. The recent decade, marked by the emergence of advanced imaging and ranging sensors and an 80-fold boost in AI computation efficiency, has ushered in a new era of perception systems capable of understanding the world around us. In this 2024 Embedded Vision Summit presentation, Vaibhav Ghadiok, Chief Technology Officer of Hayden AI, delves into essential insights gained from developing and deploying vision-based AI systems in the wild while addressing some common pitfalls and misbeliefs. He introduces “Ten Commandments” for creating an embedded vision AI product, ranging from respecting multimodality to not taking the name of (generative) AI in vain. Finally, he emphasizes the importance of an end-to-end system development approach encompassing AI, compute and a multimodal sensing suite.

Practical Strategies for Successful Implementation and Deployment of AI-based Solutions

AI models that produce accurate results on test data are a necessary component of successful applications, but by themselves they are insufficient. In this 2024 Embedded Vision Summit talk, Ritesh Agarwal, Computer Vision Lead at Globus Medical, delves into often-overlooked, but critical, elements required to create and sustain a robust application solution based on AI models. Agarwal explores the need for deep understanding of model performance and highlights techniques for identification of specific hyperparameters that can be tuned to optimize real-world accuracy. He also examines the importance of pre- and post-processing and the advantages of using multiple models in concert to improve accuracy. Finally, at the organization level, he provides recommendations for improving product development via synergy between data engineers, application domain experts, machine learning engineers, data scientists, quality assurance engineers, DevOps specialists and software engineers.

OBJECT RECOGNITION AND TRACKING

Omnilert Gun Detect: Harnessing Computer Vision to Tackle Gun Violence

In the United States in 2023, there were 658 mass shootings, and 42,996 people lost their lives to gun violence. Detecting and rapidly responding to potential and actual shootings in an automated fashion is critical to reducing these tragic figures. In 2020, Omnilert, a pioneer in emergency notification systems, launched Omnilert Gun Detect, an AI-powered platform that combines gun detection, verification, activation of security systems and notification. In this 2024 Embedded Vision Summit talk, Chad Green, Director of Artificial Intelligence at Omnilert, describes the development of Omnilert Gun Detect. He covers why computer vision is the right solution to this problem, how Omnilert went about building the product and the business and technical challenges overcome along the way. He also talks about Omnilert’s market traction, and concludes with lessons learned in building this important system.

Multi-object Tracking Systems

Object tracking is an essential capability in many computer vision systems, including applications in fields such as traffic control, self-driving vehicles, sports and more. In this 2024 Embedded Vision Summit presentation, Javier Berneche, Senior Machine Learning Engineer at Tryolabs, walks through the construction of a typical multiple object tracking (MOT) algorithm step by step. At each step, he identifies key challenges and explores design choices (for example, detection-based vs. detection-free approaches and online vs. offline tracking). Berneche discusses available off-the-shelf MOT algorithms and open-source libraries. He also identifies areas where current MOT algorithms fall short. And he introduces metrics and benchmarks commonly used to evaluate MOT solutions.

UPCOMING INDUSTRY EVENTS

Embedded Vision Summit: May 20-22, 2025, Santa Clara, California

More Events

FEATURED NEWS

MIPS’ New Atlas Product Suite Brings Real-time Intelligence to Physical AI Platforms

Arm Drives Next-generation Performance for the IoT with Its First Armv9 Edge AI Platform

AMD Unveils Its Next-generation AMD RDNA 4 Architecture with Radeon RX 9000 Series Graphics Cards

An Upcoming In-person Event from SEMI Explores Sustainable AI Systems

e-con Systems Launches a Sony IMX900-based Global Shutter HDR USB Camera Powered by the TintE ISP

More News