DeltaX has been recognized by APAC CIOOutlook Magazine as the recipient of “Top Prominent Artificial Intelligence Solutions Companies,” based on our proprietary methodology, reflecting its position in the industry. This profile has been developed by the APAC CIOOutlook research and editorial team based on insights from an interview with Stephen Kim.

DeltaX
Breaking New Grounds in Machine Learning

Stephen Kim, CEO

The rising importance of autonomous vehicles is bringing advanced driver-assistance systems (ADAS) into the mainstream. In light of the increasing prevalence of self-driving cars, manufacturers are simplifying the incorporation of computer vision and adopting newer camera setups to enhance their products.

Capitalizing on these emerging opportunities in the South Korean computer vision industry is DeltaX. Primarily focused on data management and machine vision, the company develops industry-leading capabilities for automobile manufacturers by conducting various research projects.

Adopting an Industry 4.0 approach, DeltaX answers one of Korean automobile manufacturers’ most asked questions, “Why computer vision?” with an intuitive technology for faster data collection that uses an in-house machine perception model for minimizing errors.

“Car manufacturers typically overload vehicles with sensors, but the era of smart cars is more about understanding who the users are and what they want. With this insight, we formulate new-age deep learning models to modernize the traditional mindset,” says Stephen Kim, CEO of DeltaX.

The global automobile industry is experimenting with graphics processing units (GPUs) for accelerated 3D rendering of physical spaces. They are trying to integrate LiDAR technology for tracking obstacles and enabling real-time hazard analysis. This adoption is hindered by computational limitations and the high cost associated with building a digital microenvironment within cars.

Flipping this script is DeltaX. The company makes syncing LiDAR capabilities into a vehicle’s infrastructure easy and cost-effective with a deep learning model that simplifies 3D data consolidation using a monocular camera setup.

It involves using a single camera to create 3D point clouds that have benefits beyond autonomous driving.

Dubbing the solution ’Pseudo LiDAR,’ DeltaX combines the power of the monocular setup with a powerful AI-based estimation model to facilitate the 3D mapping of captured media without the need for LiDAR. This enables faster and inexpensive operations even in areas with poor GPS connectivity, which sets the company apart from its market contemporaries.

The company installs a single camera on a car’s rearview mirror to eliminate the need for GPS. This allows users to localize or create maps based on personal routes for a truly customized experience. The technology has a gesture recognition model that collects and verifies passenger details like age, gender, behavior, and adherence to driving rules. Its AI algorithm works like a human’s short-term memory to empower smart cars with accurate driving and safety instructions.

Weighing in on DeltaX’s exceptional research capabilities are business owners seeking camera-based solutions over optics technologies. Simplifying 3D data collection using a single camera provides a cost-effective and efficient solution that outshines traditional inspection systems.

Car manufacturers typically overload vehicles with sensors, but the era of smart cars is more about understanding who the users are and what they want. With this insight, we formulate new-age deep learning models to modernize the traditional mindset

A prominent example of the proficiency DeltaX’s computer vision technology showcases is its use in automotive perception systems across automobiles manufactured by world-renowned brands. Addressing their concern about installing cutting-edge analysis capabilities in compact computation environments, the company integrated an all-in-cabin smart camera setup powered by a team operating at peak creative synergy. Initially, their inspection model needed an upgrade due to a slow and random sampling approach, which took 10 minutes per test. In contrast, the high-speed cameras from DeltaX could perform rapid and comprehensive inspections in seconds, with a 100% in-line process for all samples. It also continued providing new demos to transform their inspection capabilities.

This new automobile perception model enhanced their graphic inspection process and delivered superior precision, allowing clients to access enhanced detailing and control costs effectively.

Standing out in the Korean market with experiential diversity powered by an internationally inclusive team, DeltaX enables optimized models that adapt to the dynamic specifications of automobile businesses worldwide.

Foreseeing increased market adoption of mono-camera-based automotive perception systems and brand popularity, Kim plans to double his workforce and extend the company’s footprint to the U.S. and India. The resulting diversity will help DeltaX build robust capabilities, improve datasets, and introduce untapped creativity, setting the stage for a new era in smart cars and machine vision.