SOINN List of Products and Services

"Switch Vision V-1" is a high-precision image inspection AI that can operate on lightweight edge devices, featuring a variety of functions including unsupervised, semi-supervised, and supervised learning. Thanks to its unique algorithm, the detection accuracy is comparable to that of Deep Learning, while significantly reducing the costs associated with preparing and annotating vast amounts of data. By minimizing computational load, it can perform both learning and inference on commercially available PC CPUs, achieving high-precision judgments with inference times of around tens to 100 milliseconds per image. 【Features】 ■ High-precision identification from small amounts of data without hassle ■ Average learning time of a few minutes to several tens of minutes on commercially available PC CPUs ■ Adjust and retrain the AI while verifying on-site ■ Learning from hundreds of images on a standard PC ■ Unique algorithm different from Deep Learning *For more details, please refer to the PDF document or feel free to contact us.

• Image Processing Software
• Image analysis software

We have started selling a forecasting AI module that has a proven track record with major companies in the fields of disaster prediction and energy demand forecasting. It allows for lightweight computation, enabling learning and operation on standard CPUs. With automatic noise removal from data and automatic adjustment of learning parameters, it requires no hassle. Additionally, it can be executed in any environment, whether edge PCs, servers, or the cloud, and allows for the repurposing of trained AI models. 【Features】 ■ Customers can perform learning, forecasting, and additional learning on their side. ■ Lightweight computation enables learning and operation on standard CPUs. ■ Automatic noise removal from data and automatic adjustment of learning parameters make it hassle-free. ■ Utilizing commercially available equipment keeps both implementation and running costs low. ■ Can be executed in any environment, whether edge PCs, servers, or the cloud. *For more details, please refer to the PDF document or feel free to contact us.

• Other information systems

We have started selling an AI module for anomaly detection and predictive maintenance, which has been highly regarded by global companies in Europe. It is very lightweight in terms of computation and can operate not only on the cloud and general-purpose PCs but also on microcontrollers, Raspberry Pi, and more. All additional learning and operations can be performed by the customer. If you have any questions or inquiries, please feel free to contact us. 【Features】 ■ Direct input of time-series sensor data ■ High-speed operation ■ Can be implemented on microcontrollers, Raspberry Pi, etc. ■ Starts with normal learning ■ Capable of identifying and presenting abnormal parameters *For more details, please refer to the PDF document or feel free to contact us.

• Other information systems

We have started selling an EMS (Energy Management System) AI package that has multiple delivery and operational achievements with major industry players. Even in large-scale systems with about 20 heat source units, it is possible to perform demand forecasting and optimal control of equipment with on-site learning and real-time operation using commercially available PCs. Additional learning can be conducted on-site, allowing for daily learning and growth during operation, thereby increasing accuracy. Furthermore, everything from learning to operation can be managed by the customer. 【Features】 ■ On-site learning and real-time operation possible with commercially available PCs ■ AI operation can start in about one month if system configuration information is provided ■ Initial learning can be done with actual operational data from a few days to about a week ■ Additional learning can be conducted on-site, allowing for daily learning and growth during operation to enhance accuracy ■ Everything from learning to operation can be managed by the customer *For more details, please refer to the PDF document or feel free to contact us.

• Other information systems