SECO today announced the latest expansion of its Application Hub, introducing new AI applications that cover a broad spectrum of real-world use cases, ranging from speaker identification in security-critical environments to predictive analytics for tunnel boring machines and gesture recognition for more intuitive interfaces. By delivering these turnkey AI solutions, SECO enables client teams to focus on product differentiation while relying on pre-validated, production-ready assets.
The new application release includes:
- Speaker Identification with ECAPA-TDNN: An AI application leveraging the ECAPA-TDNN model to generate highly robust speaker embeddings. This enables accurate identification and verification of speakers in real-world environments, supporting advanced voice-based security, personalisation, and analytics.
- Object detection with RF-DETR: An advanced object detection solution based on the RF-DETR architecture, optimised for SECO Edge platforms to provide efficient, accurate, and reliable performance in real-time industrial scenarios.
- Hand Gesture Recognition: A gesture recognition application powered by advanced deep learning models, such as 3D CNNs, designed to interpret human hand gestures over time. This enables more accurate and natural human–machine interaction, especially in dynamic, real-world environments.
- Mini Agent on the Edge: A lightweight conversational AI agent designed to run efficiently on SECO edge devices, enabling localised decision-making and natural user interaction without reliance on the cloud. The agent can query sensor values in real time, control devices by turning them on or off through natural commands and interact with a local database using natural language to retrieve historical sensor data or reports without writing queries.
- ResNet – Training Experience on Device: A customizable training framework leveraging the ResNet architecture for image classification tasks. It allows developers to train and validate models directly on their target devices.
- Forecast – Settlement Predictions for TBM: A predictive AI solution for tunnel boring machine (TBM) operations, forecasting ground settlement to improve safety and support more reliable project planning.
“Our vision is to remove the barriers that typically slow down AI adoption at the edge,” says Fausto Di Segni, Head of IoT and AI at SECO. “With a growing catalogue of ready-to-deploy applications, combined with the Clea Framework and the resources of our Developer Center, we help companies move AI projects smoothly into production. This approach accelerates time-to-market while ensuring alignment with the platforms and ecosystems our clients already use.”
Each application can be deployed with confidence across platforms from Intel, Qualcomm, NXP, AMD, Rockchip, and MediaTek, backed by SECO’s enterprise Long-Term Support and validated within SECO’s hardware ecosystem to ensure operational stability and seamless integration. Developers can also benefit from the SECO Developer Center, which provides deployment guidelines, input/output specifications, and testing procedures. In addition, a free 90-day trial of the Clea Framework can be requested.
Developers can now explore the new applications directly on the SECO Application Hub.
