This is a project to help get you started with the Rayprus CSV26 camera, powered by Sony's IMX500, and AITRIOS Developer Edition. We will set up the AI camera hardware, and create a custom object detection model we can deploy to it using AITRIOS Developer Edition. Then we can run the AI model, and verify the end-to-end workflow by collecting the inference results, output as metadata, for further analysis.
By using AITRIOS and the CSV26 edge device, powered by Sony's IMX500 Intelligent Vision Sensor, you can do all of this without specialist knowledge of machine learning.
This project is intended mainly for anyone interested in AITRIOS or planning to get hands-on experience.
What is AITRIOS?"AITRIOS - The edge AI sensing platform for vision. AITRIOS combines powerful AI development tools, training services, and edge devices into an integrated Vision AI platform, empowering teams of all skill levels to build, deploy, and operate vision AI applications at the edge."Source: https://www.aitrios.sony-semicon.com/
You can get the overall picture of the AITRIOS platform from the AITRIOS website.
Traditional vision AI systems work by having network-connected cameras capture images and transmit them to the cloud for analysis. AITRIOS takes a different approach by processing AI models directly on edge devices that feature Sony's IMX500 intelligent vision sensor. With AI inference happening right on the device itself, you can upload just the results, without sending the original video footage. This eliminates the need to transfer large amounts of data, and provides an added layer of protection for sensitive information.
Traditional cameras convert light signals into image files. In contrast, AITRIOS-compatible edge devices transform the same light signals directly into data representing image recognition results.
PrerequisitesPlease note that currently (as of December 2025), AITRIOS services are available exclusively for corporate customers. We appreciate your understanding.
If you are completely new to AITRIOS, then you will need to start with the following tasks before following this project:
Register for an account on AITRIOS Portal
Go to AITRIOS Portal
- Follow these instructions to register an account.
Purchase a compatible edge AI device
- Explore compatible edge AI devices:
- This project uses the Rayprus CSV26 camera.
- For the Rayprus CSV26 camera, a PoE-compatible switching hub is required.
Register a new project on AITRIOS Portal
- Go to AITRIOS Portal
- Follow these instructions to register a new project
Purchase Console Developer Edition
- Go to AITRIOS Portal
- This project uses Console Developer Edition
- Follow these instructions to purchase a product
The workflow for the project is:
- Register the edge device to Console
- Update the device firmware to the latest version
- Create a custom object detection AI model
- Prepare the object detection edge app
- Deploy the AI model and edge app to the edge AI device
- Start inference
- Collect the inference results
I've split the project into two parts, and this first part covers the first two steps, to set up a new edge device on Console. The second part of the project is here:
Generally, programming experience isn't required. However, please note that step 7, which involves decoding and deserializing metadata, may require some familiarity with Git and Python.
Step 1: Register the edge device to ConsoleReference manual: Device Setup Guide
1. Log in to Portal and select one of your projects.
2. When the project opens, select Console from the left hand sidebar.
3. From the Console home screen, select Enroll device in the left hand sidebar.
4. Select Setup Enrollment at the top of the screen to open a dialog box.
5. Input the connection information for the NTP server and then select Generate to create a QR code for registering the device with Console.
NOTEThe CSV26 camera does not support IPv6, so it is recommended to enter the IPv4 address of the NTP server. If the NTP server name is specified and an IPv6 address is returned, connection will fail. Or, you can use an NTP server that returns an IPv4 address. I used1.jp.pool.ntp.org. Another NTP option istime.google.com. Check IPv4/v6, DNS resolution, and connectivity from terminal with:ntpdate -q time.google.com.
NOTEAlthough Managed Device -> Setup Device also provides a way to generate a QR code, it requires an additional certificate and will not provide connectivity in this case.
6. Scan the enrollment QR code with the edge device.
NOTEDisplay the enrollment QR code at approximately 15 cm per side, and scan it from a distance of about 30 cm. If the edge device doesn't read the code, you can move it closer.I expanded the QR code to fill the screen of a 14-inch laptop and held the device approximately 20 cm away, keeping it aimed at the QR code for about 10 seconds, which worked successfully more often than not.
7. When the QR code is read, the device indicator that was flashing orange will show solid red, and it will reboot.
8. You'll need to wait for about three to five minutes for the device to reconnect and register to Console. Both device indicators will show solid green.
Troubleshooting edge device connectionEven if the device indicators show solid green and solid red (Console not connected) after scanning the QR code, it might be connected to Console anyway.
Test if selecting Manage Device -> selecting the new device -> Preview changes the device indicators. If you see the camera feed in your browser, then the edge device is connected to Console.
If this doesn't work, some things to check include:
- Can a PC connected to the same PoE switch the device is using communicate externally?
- Does the edge device respond to a ping?
- Did you allow enough time after reading the QR code?
- Did you generate the QR code with other connection data besides NTP before making a firmware update?
In my experience, older firmware may be less stable, so try to register the device with only the NTP server information before making any firmware updates. You can always update the connection information later.
- Perhaps as a last resort, try using a switching hub that supports port mirroring and Wireshark (or similar) to check what kind of communications the edge device is making. This can help you understand if name resolution is failing or if there is no communication at all.
Reference manual: Device Setup Guide
The next thing we need to do is to update the device firmware. Don't skip this step; it makes the device much more stable.
TIPAlways update the device firmware on a new device or after performing a factory reset.Get the latest firmware versions
1. Go to edge device firmware downloads for Developer Edition on the AITRIOS Developers website.
2. Download the latest firmware version for CSV26 and extract the files.
3. Check the latest version information by opening the file Version.txt.
As an example, Version.txt file includes a table like:
** Version **
+--------------+----------+-----------------------+
| File | Version | UI Naming |
+--------------+----------+-----------------------+
| ota.bin | 070102PD | Main Chip |
| firmware.fpk | 020401 | Sensor Chip |
| loader.fpk | 020301 | Sensor Chip (loader) |
+--------------+----------+-----------------------+4. Check if this version is already registered for the project in Console under Settings -> Firmware.
NOTE: Do not select the Firmware v2 tab. This firmware is not compatible
5. If the latest firmware is already registered in Console, as shown above, then you don't need to register it again and you can skip to Create a new deployment config.
6. If the latest firmware is not registered in Console, then you can import it now. See Import new firmware to Console.
Import new firmware to ConsoleIf the latest firmware is already registered in Console, then you don't need to register it again and you can skip to Create a new deployment config.
1. In Console, go to Settings -> Firmware, and then select Import at the bottom of the Firmware tab.
2. Select the firmware file, select the type, and enter the version number.
NOTEFill in the version number as it appears in the file Version.txt. If you enter the wrong version number, the deployment can loop.3. You can optionally enter your own description. Finally, select Create to add the firmware to Console.
Create a new deployment configA deployment config specifies an AI model and firmware that we can deploy together to edge devices. Let's create a new deployment config for the updated firmware.
1. In Console, go to Deploy to Device.
2. From the Config tab, select New Config at the top of the screen.
NOTE: Do not select the Config v2 tab. This firmware is not compatible.
3. Enter a name for the config, and you can optionally add a description. Select "None" for AI Model, because we only need to update the firmware, and then select the new firmware versions for Sensor Chip, Main Chip, and Sensor Chip (loader).
4. Select Create and confirm that the new deployment config is registered in the list.
Deploy new firmware to the edge device1. In Console, from Deploy to Device -> Config tab, select the new deployment config we just created.
2. From the panel that opens on the right side of the screen, select Deploy.
3. Finally, select the new edge device to deploy our firmware update config on. Deployment begins.
You can check the deployment progress in Console under Manage Device. Select the new edge device, and in the panel on the right side of the screen, expand the area under Status -> Deployment.
Deployment is done when the value for Status is "success".
TIP: The status does not update automatically. Select Reload to refresh the status.When in trouble
If you encounter any issues while following this article, comment on the article. Please note that it may take some time to respond to comments.
When I faced issues that I couldn't solve during device registration, I was able to get past them by pressing and holding the reset button for 30 seconds to perform a factory reset, and then repeating the same steps, so please try that if you're really stuck.
Additionally, if you have any issues regarding AITRIOS that are outside the scope of this article, try the AITRIOS Support site,and if you still can't find the solution there, contact us from the form at Contact Us.
ConclusionIn this article, we registered a new edge device in Console and updated the firmware to the latest version.
In the second part of the project, we will create and deploy an AI model to the edge device, perform inference, and collect the inference results, so please make sure to check that out as well!
Comments