Here's a potential way to use adversarial techniques to generate training examples that could improve the accuracy of this approach: https://twitter.com/RoboTed

Here's a potential way to use adversarial techniques to generate training examples that could improve the accuracy of this approach: https://twitter.com/RoboTeddy/status/1228828411050655744 https://twitter.com/RoboTeddy/status/1228828411050655744