DeepX and Lotte Innovate have embarked on a collaboration to deploy domestically produced AI semiconductors in actual industrial sites. Moving beyond simple technology verification, the two companies are expanding the scope of their cooperation to mass production, applying AI chips to operational sites such as roads and retail stores. Starting with this, they plan to extend the application areas to the manufacturing, robotics, and safety industries.

In this announcement, the two companies stated that they have entered into discussions for the commercialization phase after completing the proof-of-concept (PoC) of the DeepX NPU-based solution they have been conducting. The starting point for the collaboration is transportation infrastructure and retail stores. The priority areas of consideration are applying DeepX chips to AI edge cameras installed in high-traffic areas of highways and to smart CCTVs in large stores.

In the transportation sector, the key is a structure that performs vehicle recognition and anomaly detection directly on-site without going through cloud servers. This so-called on-device AI approach is explained as reducing communication burdens and enhancing real-time responsiveness by eliminating the need to continuously transmit video data to a central server. This has the potential to lead to a demonstration case of an intelligent transportation system that integrates edge AI into public infrastructure.

In the retail sector, plans are being pursued to apply AI to CCTV video analysis within large stores to handle customer movement tracking, safety management, and inventory monitoring. The goal is to perform AI inference on on-site terminals while reducing power consumption compared to server-centric methods in environments where thousands of cameras are in operation. Through this, both companies plan to jointly assess store operational efficiency and management systems.

The division of roles has also been concretized. Lotte Innovate will provide industrial data and operational experience, and utilize its own business sites as testing grounds. At the same time, it plans to accelerate the spread to Lotte affiliate sites by establishing a standard development procedure to optimize its proprietary AI algorithms for DeepX's software development kit, DXNN.

DeepX is responsible for hardware optimization and technical support to ensure stable operation in low-power and low-heat environments. The scope of support also includes the deployment of dedicated personnel, technical consulting, and supply chain management. The two companies plan to expand this collaboration beyond the transportation and retail sectors to manufacturing, robotics, smart factories, and safety control, aiming to accumulate application cases for edge AI semiconductors in domestic industrial sites.