OSD32MP1 Powers AI at the Edge – MPCAM

When most of us think of Artificial Intelligence (AI) we think of large server farms doing complex calculations.  In this paradigm for a product to leverage AI it would need to connect and send data to these servers to do their AI magic and then return the answer to the device, similar to Alexa.  However, with the continued advancements in performance along with the reduction of size and power of processors, like the OSD32MP1 System-in-Package (SiP) featuring the STM32MP1, it is possible to run AI algorithms right on the local product, at the Edge.

The MPCAM from Siana Systems is an example of just such a platform.  Featuring the OSD32MP1 SiP it is a platform that does Computer Vision (CV) at the edge.

AI at the Edge

Being able to run AI applications at the edge has some big advantages, especially when talking about computer vision.  Computer vision requires a video stream, which could produce a lot of data to do the analysis work.  If you are running your CV algorithms on a server farm then you need a fast, reliable connection to the servers to process that data and return the answers.  When running CV at the Edge, you don’t need an always-on fast connection to the server farm.  This allows the device to operate in areas where connections might not be reliable or are cost prohibitive to implement.

Another advantage in AI at the edge is increased privacy for the user.  As we described above in the server based AI paradigm, the user data, in this case video, needs to be sent to the servers.  Once the data has left the device and resides on the servers there is no telling what could be done with it, as you can image there are plenty of privacy concerns.  With Edge AI, the raw data never has to leave the device for the AI to work.  Your users can rest at ease knowing that their videos and personal information are not being shared and stored in the cloud where they no longer have control of it.

MPCAM

The MPCAM is a low cost platform designed to bring edge AI, specifically Computer Vision (CV), projects out of the lab and into the real world.  It allows you to quickly build and deploy prototype CV applications without having to worry about getting different development kits to work together.

At the center of the MPCAM is the OSD32MP1 System-in-Package that integrates the STM32MP1 microprocessor from STMicroelectronics, DDR Memory, Power Management, EEPROM, Oscillator, and passives in a tiny 18mm X 18mm BGA package.  The Dual Core Arm® Cortex® A7 STM32MP1 efficiently runs Linux allowing you to take advantage of many of the opensource AI frameworks.  The integration provided by the OSD32MP1 System-in-Package enables a compact platform that significantly reduces the size and risk when compared to a discrete chip down compute system.  It provides the MPCAM with a small, powerful, and efficient computing platform for its CV applications.

Accompanying the SiP is a 5MP OV5640 image sensor with S-Mount Lens, a Coral TPU Accelerator, 4GB eMMC flash, and an ethernet port.  It comes preloaded with OpenSTLinux with X-Linux-AI extensions, OpenCV 4.1, TensorFlow Lite 2.5.0, and Python 3.8.

The MPCAM leverages these frameworks to provide 3 Demos out of the box.  They are Object Detection, Pose Detection, and People Tracking.  Each of these can be used as a starting point to develop your own computer vision application.  You can learn more about what is included with the MPCAM, getting started information, and beginning development information on Siana Systems’ Wiki.

The combination of the hardware, frameworks, and demos provides you a great starting point to develop your first Edge CV application with plenty of room to grow.

Want to Get Started?

Are you ready to begin your edge AI journey with Computer Vision with an easy to use all in one platform?  You can order one of the MPCAMs from the Siana Systems Store.  If you are looking for something a bit more customized for your use case or want more information on the MPCAM contact Siana Systems.  Finally, if you would like to know more about how the OSD32MP1 can power your edge AI applications reach out to your local Octavo Representative.

Stay up to date on all of the latest innovations in System-in-Package and Edge AI.  Sign up for the Octavo Systems newsletter.