NewsMPCam and MPTherm, computer vision starts with an STM32MP1 in an all-in-one...

MPCam and MPTherm, computer vision starts with an STM32MP1 in an all-in-one solution

Category articles

The MPCam and the MPTherm are two development boards for STM32MP1 that are manufactured by Siana Systems, an ST Authorized Partner that demonstrate how machine learning and computer vision on the edges are becoming commonplace. Both systems share similar specifications. They are both OSD32MP15x systems-in-packages of Octavo which we discussed in a blog post from a while ago. The model that Siana picked comes with the 512 MB DDR3. The boards come with 4GB of eMMC flash that can store images that can be booted and the SD slot as well as it has an Ethernet port USB host capability, and a 9-volt connector to connect connecting an adapter to power externally. The MPCam comes with a 5-megapixel image sensor, and the MPTherm has an 80×80 thermal sensor by Lynred.

MPTherm and MPCam, or the best way to get started with computer vision

What makes Computer Vision difficult?

Computer vision remains a challenging technical challenge to address. However, more and more businesses with no engineering expertise would like to reap the benefits of it. Retail stores would like to employ people-tracking apps to boost efficiency while protecting the privacy of customers. Malls require better methods to track capacity, particularly after a worldwide pandemic. From social distancing to simple recognition of objects Computer vision is gaining increasing focus. As Siana Systems shared, the issue is that nearly 90% of the projects do not pass through the prototype stage. Companies don’t know the requirements they require or the requirements for commercialization. Therefore it is necessary to create development boards that help enable managers to return to their reality.

What exactly does computer vision look like?

One of the most difficult tasks is to understand what a computer vision device appears to be. MPTherm and MPCam assist companies by directly placing an extremely small PCB in the hands of decision makers and developers who can easily see what they’re working with. It’s the reason that Siana has chosen to work with the Octavo SiP instead of starting completely from scratch. The company has shared that this strategy will result in a speedier time to market and smaller PCB. Although she had the skills, Siana wanted to focus on the application instead of optimizing the RAM, devising power management, or any of the other technical applications required by a microprocessor.

What is the reason for developing computer vision software on the edge?

Businesses also have issues with the application. Many people don’t even think about sending video streams to the cloud, but fail to consider the infrastructure and costs that are required to implement this strategy. For example, a mall that relies on Wi-Fi will soon find its conventional access points slow down when it was streaming numerous video channels throughout the day. This means that edge devices that offer AI capabilities is typically the most crucial for a successful software. It’s why the MPTherm and MPCam use an STM32MP1 along with a Coral, which is a tensor processing device (TPU). In reality, the device must be small , yet sufficient to be able to process calculations locally, which will reduce the bandwidth requirements.

How do you create computer vision software on the edge?

Although those who are aware of the necessity of AI in the edges may not know what is involved in an Linux operating system, and how to design the ideal user experience that will aid in the management of system processes and other functions. Therefore, MPTherm and MPCam come with the ability to boot a Linux image that is based on OpenSTLinux which is an embedded Linux distribution. Furthermore, Siana worked closely with ST to implement our XLINUX-AI software, which allows you to run three demonstrations right out right out of the box. X-LINUX AI includes Python 3.10, TensorFlow Lite 2.8.0, OpenCV 4.5, and much more.

The frameworks allow Siana to offer image recognition and detection of objects. In addition, Siana provides a unique people-tracking software together and People tracking is a particular challenge since it’s very easy to produce inaccurate results when you count the same person several times or not being able to track their movements. In reality, it’s not enough to have the correct hardware or OS and framework. Every use case requires new algorithms, datasets and optimizations. Through the use of OpenSTLinux and X-LINUX AI, Siana helps customers get to the point of starting. The MPCam and MPTherm come with an internet server. Users can connect to it via a browser and start using the demo apps in minutes.

How do we promote computer vision on the edges?

The boards made by Siana provide a unique chance for businesses to experience computer vision without the need for an in-house data scientist. The participant in ST Partner Program ST Partner Program explains that it’s typically the best way to assist companies in determining whether computer vision is suitable for them and what is that it would require to make it real. The end product is similar to the final layout of PCB and will help you create an application specifically tailored to meet your requirements. Utilizing an STM32MP1 siP and Stura’s solutions for software, Siana demystified computer vision for companies that have little technical experience to help them make the right choice for their company.

What are the main challenges in deployment?

It’s the ST Authorized Partner is now seeking to solve the next hurdle that is deployment. Many times, the installation of boards can be a challenge, particularly in cases where the customer does not have the necessary technical expertise. Siana is therefore creating kits to make field deployment simpler. Siana is also focused on microcontrollers. In fact, even though the company has no details yet however, it has stated that it’s working more on embedded systems that use the STM32H7 or STM32F7and that advances regarding performance as well as UI design and speed are driving ever more companies to embrace microcontrollers. The member of the ST Partner Program is also a TouchGFX implementationer to support this very reason.

Siana and TouchGFX and the next generation of embedded systems

What are the benefits of working with the STM32MP1?

Although the MPTherm and MPCam solely use ST’s microprocessors, Siana shared that they’ve also been working on other projects that utilize STM32 microcontrollers. They were also the first to adopt TouchGFX. It was therefore interesting to see how the company works with MPUs and other MCUs. Indeed, working with microprocessors as well as Linux allows the creation of UX and user experiences simple. The company has even an UX specialist who specializes in graphic design and not developing code. With a framework such as Crank which is an additional ST Partner Program member the designers do not require knowledge of code generation. It simplifies workflows and facilitates communication with customers.

What is the reason why TouchGFX getting more and more well-known?

However, Siana also shared how they’re witnessing similar trends in STM32 MCUs. Their teams heavily rely in their TouchGFX simulator. It allows the designers to collaborate together with their graphic teams or customers without having to deal with the codebase. The company also stated the performance of devices like the STM32H7 bridge the gap between microprocessors and microprocessors that allow engineers to do more with microcontrollers. There are some noticeable variations between UI style among Linux or a real-time application. Additionally, some applications require the power provided by an MPU. We’re not saying they can be interchanged. however, it’s interesting to note that a design firm that has a vast knowledge of embedded systems based on Linux is increasingly investing in MCUs and TouchGFX.

As computer vision on the edge becomes commonplace We expect the markets to shift towards cost-effective solutions that will invariably include microcontrollers. It is interesting to observe how the efforts the TouchGFX teams have done in the reduction of tension and making the framework user-friendly directly affects the work of design houses such as Siana. In the event that computer vision has already become in the realm of possibility, advances like these will make edge devices all over the place. That’s also the draw we made at electronica 2022 when the MPCam became quite popular at our ST Booth.

Michal Pukala
Electronics and Telecommunications engineer with Electro-energetics Master degree graduation. Lightning designer experienced engineer. Currently working in IT industry.