NVIDIA SBC Jetson Nano Version: B01 Development Kit, Quad - Core ARM A57, 4GB 128 - Core NVIDIA Maxwell GPU, 64-Bit LPDDR4

Fantastic entry level device for GPU usage. Wouldn't replace your PC with it but there is nothing else like it on the market. In terms of running YOLO on it like the previous reviewer said, for starters one just allocate more swap memory and second of all that is not an edge detection algorithm. Nvidia have great tutorial content and you can definitely run real time inference (see https://github.com/dusty-nv/jetson-inference for a fantastic starter to object detection). Fantastic entry level device for GPU usage. Wouldn't replace your PC with it but there is nothing else like it on the market. In terms of running YOLO on it like the previous reviewer said, for starters one just allocate more swap memory and second of all that is not an edge detection algorithm. Nvidia have great tutorial content and you can definitely run real time inference (see https://github.com/dusty-nv/jetson-inference for a fantastic starter to object detection).

What are most people going to do with an AI edge device? Probably run some sort of object detection. Currently the most popular object detector is Yolo. Do any of the Nvidia Jetson Nano tutorials use Yolo?, Nope. Guess why: Some Yolo models after they are converted with TensorRT, don't fit in the Nano's 4GB of RAM. Others that do fit, run like a dog i.e. 2FPS. What are most people going to do with an AI edge device? Probably run some sort of object detection. Currently the most popular object detector is Yolo. Do any of the Nvidia Jetson Nano tutorials use Yolo?, Nope. Guess why: Some Yolo models after they are converted with TensorRT, don't fit in the Nano's 4GB of RAM. Others that do fit, run like a dog i.e. 2FPS.

excellent development board for Ai work excellent development board for Ai work