The safety and security of public (places and) facilities is a serious concern that inspired us to develop an AI that delivers world-class accuracy in recognizing individuals. Even in operating conditions of varying camera angles and changing ambient light, the system maintains a high level of recognition accuracy. This has allowed us to refine technology that can identify and track specific people as they move through places and are captured across multiple cameras. 

Person re-identification between cameras
Face recognition
 
2-stage Auto Encoder


Power and industrial plants want to maximize uptime. We are realizing this with an AI system that delivers highly precise, early detection of anomalies in large-scale plants. A two-stage autoencoder identifies and learns complex relationships within the large volumes of time-series data collected from sensors installed throughout the plant. It can then detect early signs of abnormality that are normally overlooked during routine operational changes, bringing increased efficiency to plant operation and maintenance.
More details →

Related Links:AI Technology Catalog

2-stage Auto Encoder

Our electricity market trading optimization technology supports strategic trading by energy providers. The technology uses our proprietary high-precision forecasting of renewable energy output and market prices to calculate the optimal bid volume for the spot market, and determines both the timing and volume of bids in the intraday market. This enables decision-making that maximizes revenue while accounting for risk.

Related Links:AI Technology Catalog

In the design of things like advanced materials or high-performance power semiconductors, the large number of design parameters makes manual tuning increasingly difficult. 
We are working on an AI-based, high-dimensional Bayesian optimization technology that automates the process. It is advancing data-driven design, raising development efficiency, and realizing high-performance products in a range of fields. 
More details →

Related Links:AI Technology Catalog

We are conducting research and development of development methods that utilize generative AI with the aim of improving the efficiency of the software development process. Using generative AI based on a large volume of software assets, the necessary tasks are performed automatically while passing on know-how at each step of the development process. By combining the inheritance of know-how and productivity improvement through automation, we support social infrastructure by maintaining highly reliable software.

We are developing a proprietary agentic AI system that generates software with user-specified functionality by engaging in dialogue to interpret ambiguous and complex user requests. The AI uses this interactive process to analyze and break down problems, select the most suitable technologies from a broad array of assets, including various AI techniques, and automatically generate and implement the software code. This approach enhances the efficiency of software development and helps address a wide range of on-site challenges.

Related Links:AI Technology Catalog

As skilled people age out of the workforce, it can be hard to replace them. To counter this, we are developing an AI that learns and applies evaluation processes that skilled inspectors use to detect defects in ultrasonic images. By delivering as a web-based service, we aim to overcome shortages of skilled engineers and maintain inspection quality. 

AI-Powered Inspection Data Evaluation