09-15, 12:30–13:00 (Europe/London), Assembly Room
An introduction to the Jetson Nano Developer kit by a data scientist
This talk will describe my journey of experimentation with the Jetson Nano Developer Kit, which is a low cost, single board computer about the size of the Raspberry Pi with super machine learning powers onboard.
As an industrial engineer and data scientist with limited "small factor" computing experience, I'll explain how I got on getting it set up, with my review of the spec and features of this device. Once all running and with the correct dependencies, we will move on to how to start training and inference with deep learning models in Python and both Tensorflow and PyTorch, followed by use of Jetson.GPIO to interface with micro sensing units.
After introducing the device, suitable peripherals and its configuration steps, the talk will include a live demo showing the sensing and inference capabilities and also will report on the outcome of a couple of home projects where I deployed the Jetson Nano and watched it learn, infer and interact with the real world.
I'll conclude with a summary on this type of "edge machine learning" device- and my view on whether this new distributed computing paradigm will represent a threat to cloud-based machine learning in the medium to long term.
Frank is a data scientist, recovering engineer and Python user working on smart city related projects including most recently, road network condition modelling and eco-friendly car journey planning. He manages a team of data scientists based in Amsterdam whilst working remotely from Bristol where he also helps run the local PyData meet up.