Robert Bond, an NVIDIA engineer, created an “ant annoyer” using deep learning and a Jetson TK1 developer kit to pinpoint ants scuttling across his kitchen floor and target them with a 5 milliwatt laser beam.
Robert mentioned the project was not even practical before he got his hands on the Jetson board, which puts unprecedented computing performance in a low-power, portable and fully programmable package, plus it has support for the Caffe deep-learning software development framework.
He trained a neural network in nearly 30 minutes using his personal home video of ants to recognize the pesky insects.
“Deep learning has a lot of potential,” Robert says. “And I was really impressed that the Jetson had the horsepower to do all the processing needed to do this in real time.”