While at re:Invent, I got to take a deep learning workshop to learn about the new capabilities of AWS such as SageMaker and Greengrass. We used a new device created by AWS and Intel called DeepLens to build an image classification model, deploy it to the device, and use the model to predict image labels. In reference to the TV show Silicon Valley, we trained a neural network to determine if the image from the video camera contained a “Hotdog or Not hotdog.” My machine learning model performed better when the hot dog had mustard on it.
Fun aside, the DeepLens device is a really powerful way to learn image and video-based machine learning and artificial intelligence. Once you get ahold of the device, you can follow the workshop by downloading the materials and code from GitHub.