“This was bought for me as an xmas present. Installing the hardware and software was relatively easy.
Some of the rpicam post processing files were aimed at the Hailo8 not the Hailo9L but that was easy enough to fix, by editing the json post processing files, and I got them running. The rpi5_examples worked very well, especially the pose estimation using the video of people walking across the crossing. I also used a dowloaded video from my dashcam to detect vehicles - nice.
Most of the models are trained on American ojects. E.g. large cars are identified as trucks.
Now the downside. Models in the hailo model zoo etc seem to be for Hailo8 not Hailo8L. Links to the 8L hef files are broken. That means you need to recompile them using the Hailo DFC (which requires a high spec pc) My quibble is about the lack of support for the 8L. After all, most of us are just dipping our toes into this , we need to see working examples for the 8L.
The lack of support for 8L is the reason for 3 stars.”
“exactly as described, planned for a AI based project where i required an SSD too, so will definitley fit the bill. delivered quickly by the Pi Hut too.”
“The official AI Kit had been out of stock for a while when I bought this. I have another Pi 5 with an NVMe drive and so purchasing this was a no-brainer. The trade-off with having both an NVMe and the Hailo 8L means gen-3 speeds aren't possible but I haven't really noticed any significant performance difference. I have a 2TB NVMe and the Hailo 8L using this HAT and I am very happy. Thanks!”
“So much easier to use than the Google Coral TPU. Plugged in, it worked no tricky configuration changes. The HailoRT library is so easy to use. Already have a process discarding videos that do not have any birds in them from my wildlife camera. Highly recommend it.”
“Extremely easy to set up, documentation is top notch and the examples are informative, as usual from the raspberry pi team. Looking forward to seeing what the NPU can run, a great addition to my robotics toolset.”
