So i love the roboflow software and has been super helpfull to me in creating models and deploying them but now i wanted to run my model offline, and in my rasberry pi 3 it is unable to download the inference_sdk library i suppose this is due to ram issues i upgraded to rasberry 4 and later 5 and it works on both so for people who want to use rasberry pi i would suggest minimum 4. now to fix this issue on pi 3 i decided to run it using yolo, i downloaded the data set from roboflow and trained my own custom yolo model which actually works surprisingly well and doesnt required internet, i just wanted to share this with others incase anyone wants to deploy models locally after training them
Edit: this is also fully free to use and doesnt consume credits on roboflow
Hi @Subhranshu_Mohanty - that’s the path we recommend for devices like the Pi where we don’t have existing deployment support. Basic and Growth customers are able to download weights to make this even easier.
One thing you’re going to want to watch out for is model licensing. YOLOv8 is distributed under an AGPL-3.0 license, which requires associated code changes to be open-sourced. Our Growth plans come with a license to use the model commercially outside of the Roboflow ecosystem.
as of now i have only using this for private use for research and projects, but later on when i would require it for commercial use i would definitely require a license i suppose at that time roboflow would be the best option as i can both train and get the license