Bookmark and Share

Free Webinar Explores Developing Low-cost, Low-power, Always-on, Vision-based AI Solutions

On October 30, 2018 at 9 am PT (noon ET), Lattice Semiconductor's Deepak Boppana, Senior Director of Marketing, and Gordon Hands, Marketing Director for IP and Solutions, will present a free one-hour webinar, "Architecting Always-On, Context-Aware, On-Device AI Using Flexible Low-power FPGAs," organized by the Embedded Vision Alliance. Here's the description, from the event registration page:

Driven by concerns such as network bandwidth limitations, privacy and decision latency, interest in on-device Artificial Intelligence (AI) is increasing. Always-on context awareness is a key requirement in devices at the Edge, including mobile, smart home, smart city, and smart factory applications. Many of these Edge devices are battery-operated or have thermal constraints, leading to stringent power, size, and cost limitations. Additionally, the inferencing solution has to be flexible enough to adapt to evolving deep learning algorithms and architectures, including on-device training.

Given this unique mix of requirements for on-device Edge AI, developers need to architect their systems thoughtfully, both at the system level and the chip level. FPGAs are proving to be well suited for implementing machine learning inferencing, given their inherent parallel processing capabilities. In this webinar, Lattice Semiconductor uses its experience in developing always-on, vision-based AI solutions to illustrate these tradeoffs.

The first section of the webinar explores the pros and cons of various system-level architectures in implementing flexible, always-on AI solutions, ranging from standalone FPGAs to the combination of FPGAs and high-end AP SoCs or low-end MCUs. The focus then shifts to chip-level optimizations, with an overview of the design choices available to developers, including FPGA resources such as embedded memory and multipliers, network size, quantization and clock rate. These choices impact key metrics such as decisions per second, the ability to resolve small features, decision accuracy, power consumption, size and cost. Example designs such as face detection and human presence detection are highlighted to explore these optimizations across solutions ranging from 1mW to 1W. A 15-minute question-and-answer session will follow the presentation.

To register for this free webinar, please see the event page. For more information, please email

UPDATE: Now that the live webinar is over, see here for the on-demand archive.