Hi
I’m using raspberry Pi 5 and Hailo 8 using developer kit.
I’ve 3 cameras connected to it and I want to save the videos, since raspberry pi 5 doesn’t have video encoder support I though whether there is an option to utilize Hailo 8 card to do it. Similar to ffmpeg integration with CUDA or Vulkan support.
Thanks in advance and looking forward to your information,
Erez
Hi @Erez,
Hailo-8 doesn’t have an encoder.
Thanks for the reply.
I’m wondering what is the best possible option to analyze & record video on Raspberry Pi 5…
Welcome to the Hailo Community!
The Hailo-8 AI accelerator was designed to run inference of AI networks very efficiently at the edge. So, if you mean that with “analyze” you will enjoy using the Hailo-8 for that part of your application.
For recording the video you will need to make use of the Raspberry Pi 5 hardware. I would recommend to look at the examples and documentation from the Raspberry Pi community for that part.
We will be happy to help you with the inference part. Check out the examples already available from Raspberry and us.
Unfortunately Raspberry Pi is not offering any hardware encoder, only decoder.
see below, ffmpeg is implementing CUDA support, i’m wondering if the same can be achieved by Hailo, what is your opinion?
The Hailo-8 has been designed to accelerate inference of neural networks. Our workflow requires using the Hailo Dataflow Compiler to convert models in ONNX or TFLite format to our HEF (Hailo Executable File). Because our hardware is more specialized and not generic we do not support low-level programming.
We do have our Hailo-15 Vision Processor family that also have video encoders for H.264 and H.265. However they are not available in a M.2 form factor. You can get a System-on-Module e.g. from SolidRun and others will follow.