Mark Filion
April 10, 2024
Reading time:
With a multitude of applications, the STM32MP2 is being put to the test as an integral platform for open source machine learning at Collabora's booth at Embedded World this year. As one of the featured demos, it highlights how some neural networks can be run on the edge, enabling decisions be made without relying on the Cloud.
Using TensorFlow Lite models optimized for the STM3MP2 NPU along with an upstream-ready H.264 encoder (Video4Linux2), this demo also showcases GStreamer's all-new analytic metadata framework via ONVIF metadata streaming over RTP in Rust.
![]() |
Due to the need for an NPU that can handle analyzing video images en mass, the STM32MP2 is an ideal and affordable chip to enable edge AI solution in industrial environments. As an essential piece for machine learning functions, it can also be applied to the fields of industry 4.0, security, robotics, and real-time quality control.
Stop by Booth 404 in Hall 4 to get a firsthand look at edge intelligence on STMicroelectronics' STM32MP2!
26/03/2025
The kernel release emphasizes the continuing growth and maturity of the Linux ecosystem and delivers new features, improvements, and optimizations…
26/03/2025
Collabora is thrilled to see its partner, Hololight, join the Khronos Group and become an active member of the OpenXR Working Group. We…
11/03/2025
Using Mesa 25.1, Nouveau users will default to Zink+NVK instead of the old OpenGL driver, shifting towards an OpenGL on Vulkan.
Comments (0)
Add a Comment