Hi! I was wondering if there is more information about RF-DETR deployment anywhere? I read this on the website: “RF-DETR is developed for projects that need a model that can run high speeds with a high degree of accuracy, and often on limited compute (like on the edge or low latency).” and found this when I googled it, however I cannot seem to find that information when I click on that page. Is it really possible that it could be deployed on a Raspberry Pi or is that google search result just misleading?
Hi there, you can definitely deploy your RF-DETR model on a Raspberry Pi!
Here are a few links related to deployment in Roboflow:
Raspberry Pi does have limitations with compute, but RF-DETR is designed with this in mind!
1 Like
Hi, thank you so much for the quick reply! Do you happen to know where I can find that reference about RF-DETR’s capability to be deployed on a Raspberry Pi, so that I can cite it? Many thanks!
The best I can supply is this section: Install on Raspberry Pi - Roboflow Inference
The only models not supported are models that require a large GPU, such as Florence 2.
Thank you! Would that be for all of the models or perhaps Nano or Small?
I just noticed this on the repo, but I‘ve started training my models before this update:
2025/07/23: Released three new checkpoints for RF-DETR: Nano, Small, and Medium.
Thanks a lot once again!
All sizes should be able to run on the Pi! You will get faster performance from the smaller model!
Okay amazing, I have been using it for early wildfire smoke detection research and this is very reassuring! Thank you!