Leading AI startups from Korea attended NVIDIA’s Inception Connect Seoul
In his speech, Sung detailed how StradVision is using its lean technology to bring safe and reliable Autonomous Vehicles to our roadways.
The event, hosted by tech giant NVIDIA, drew a wide variety of attendees from NVIDIA’s Korean Inception Program — which accelerates the growth of more than 4,000 cutting-edge AI startups by offering go-to-market support, deep learning technology training, access to exclusive events and technology discounts.
The goal of the Inception Program is to nurture startups leveraging AI, advise companies when to deploy deep learning models, and offer Inception members a vast network of deep learning experts.
Founded in 2014 and co-headquartered in San Jose, Calif., and Seoul, South Korea, StradVision develops deep learning-based perception software for ADAS and Autonomous Vehicles.
StradVision CEO Junhwan Kim said that for Autonomous Vehicles to become a reality, it’s extremely helpful for companies like NVIDIA to work closely with growing contributors to the ecosystem, including StradVision.
“We are appreciative of the assistance NVIDIA is providing as we move StradVision’s innovations from the laboratory to the streets,” Kim said. “There’s no doubt that the Inception program boosts the speed at which we can bring our products to market.”
StradVision’s efforts in the AV sector revolve around its SVNet deep learning-based vision processing software.
Dr. Sung — who previously served as a Principal Engineer at Samsung S-1 and a Senior Engineer at LG Electronics — detailed how StradVision’s software focuses on high-level perception abilities — including Lane Detection, Traffic Light and Sign Detection, Object Detection and Free Space Detection. Tier 1 suppliers then take this technology and apply it to their safety functions, such as Adaptive Cruise Control.
Dr. Sung also detailed how SVNet’s network is small, yet powerful, when compared to other players in the perception software market.
He explained how StradVision’s Deep Neural Network is the only network available that meets accuracy and computational requirements for commercial use on lean automotive embedded hardware. Also part of the presentation were details on StradVision’s internal camera software, SVNet Internal, which allows for driver monitoring to improve safety.
In addition, Dr. Sung discussed NVIDIA’s CUDA Deep Neural Network (cuDNN), which allows deep-learning researchers and framework developers to create high-performance GPU acceleration. He explained how StradVision uses cuDNN to optimize SVNet functions on NVIDIA modules TX2 and PX 2.
Dr. Sung said he appreciated the chance to speak at the NVIDIA event, due to the important role the company plays in the development of today’s ADAS systems and Autonomous Vehicles.
“This event brings together the leading companies who are blazing the trail in perception technology, and I’m honored to be given the chance to present StradVision’s work,” Dr. Sung said. “With so many systems in play, making Autonomous Vehicles a reality will require collaboration between all areas of the autonomous supply chain, and we’re excited to do our part to advance this technology.”