What is the fastest way of testing the Hailo-8R on Verdin-IMX8M-plus on Ivy Carrier on RTSP streams coming from IP cameras?
Can I follow this guide? Even though it is for the mallow carrier?
https://www.toradex.com/de/blog/accelerating-edge-ai-hailo?srsltid=AfmBOortzffn7n55EpshnckgF8j3RIVkxOEfyTNMO16A7T0qyR_pHVb2
Can I omit this layer because I am using RTSP from IP?
bitbake-layers add-layer meta-toradex-framos
Thanks
Hi, I’m the Toradex AI Assistant. The Toradex Support team will follow up with you shortly!
Did you like my answer?

Hello there!
When working with the Verdin iMX8M Plus and Ivy carrier board, it’s important to consider a few compatibility aspects:
-
The Verdin iMX8M Plus paired with the Ivy carrier board supports PCIe Gen 3, which is compatible with the Hailo-8R. You can find more details in the Ivy compatibility documentation and the Verdin iMX8M Plus module documentation.
-
The guide you mentioned uses a different hardware setup, specifically the Mallow carrier board, as shown in this blog post. While the AI acceleration concepts remain similar, you’ll need to adjust the hardware setup steps for the Ivy carrier board.
-
Regarding the meta-toradex-framos layer, it’s designed for Framos camera support. Since you’re using RTSP IP cameras, this layer isn’t necessary for your setup.
However, you should still include the following essential layers in your build:
bitbake-layers add-layer meta-hailo/meta-hailo-accelerator
bitbake-layers add-layer meta-hailo/meta-hailo-libhailort
bitbake-layers add-layer meta-hailo/meta-hailo-tappas
bitbake-layers add-layer meta-hailo/meta-hailo-vpu
For more detailed setup instructions tailored to your specific hardware configuration, please refer to the official Toradex documentation.
If you need further assistance with setting up AI acceleration using RTSP streams, feel free to ask!
Best regards,
Toradex AI Assistant.
Dear @eliaspromiseq,
We have had a look at the Hailo-8R module and it appears to be making use of the mPCIe connector. We have not tested it on our side, but it will likely work. The answer given by @ToradexAI looks decent and we do not have anything to add at this point. Please let us know how you fare with this.
Best regards,
Collin