Intelligence allows machines to interpret and respond to their environment. We architect newer artificial intelligence models to play a role in various applications such as autonomous sensing, surveillance and service orchestrations.
Businesses and organizations are increasingly turning to local AI instances to gain a competitive edge. By
leveraging AI models tailored to their specific requirements, we can gain deeper insights, make our systems
aware, and optimize their processes.
One of the key benefits of local AI instances is the ability to infer patterns and insights on data
directly on the edge. No external cloud nor remote connections are needed for our AI deployment. For
example, we use edge AI for industry tailored applications, such as quality assessment, or to reveal network
traffic quality.
Another area of innovation in AI is the development of new architectures. SMA-RTY’s researchers and
engineers are developing new AI architectures that are more efficient, more accurate, and better suited to
the company’s specific inquiries. These new architectures can enable us to solve problems that were
previously impossible or too expensive to solve.
Introducing BASEN, your gateway to embedded AI processing! This GPU-compatible support card empowers seamless AI integration, boasting a multitude of sensor interfaces. Revolutionize your system's capabilities with BASEN's versatile platform, designed to harness AI potential while effortlessly integrating diverse sensor interfaces.
BASEN is an original system for integrating AI algorithms. An NVIDIA Jetson module makes up the computing core, and it is equipped with numerous interfaces dedicated to embedded systems. BASEN includes its own power regulation system, and can connect up to 5 vision systems via the MIPI protocol, 2 separate Ethernet LINKS. of course, several interfaces such as HDMI, USB2, USB3…
Given the rapid evolution of hardware architectures for image processing and AI, it was decided to adopt a modular approach to selecting the BASEN system’s computing core. As a result, BASEN can support a variety of NVIDIA System-on-Modules” (SoMs) in SO-DIMM format. The board is currently based on a XavierNX SoM, but an upgrade to a future-generation core is possible.
SMA-RTY utilizes advanced AI models combined with custom filtering techniques to enhance data fusion. In this video, our approach improves image registration methods to ensure high accuracy in aligning diverse data sources (RGB and LWIR in this case). By merging AI with precision filtering, we provide improved insights across multiple modalities."
AI-based methods shine in the field of region of interest tracking. By monitoring specific regions within a
video feed, we can automatically monitor specific areas and detect anomalies or patterns. This can be
particularly useful in security applications, where we can use AI to monitor large areas for potential
threats, e.g., smart city.
We leverage state-of-the-art object detectors, which we fine-tune to match our specific modalities. This
customization allows us to deliver high-quality, modality-specific analytics to our partners.