Original article by IZ's Summer trainee Rasmus Cornelius.
For most people artificial intelligence, machine learning and deep learning are something abstract and hard to wrap your head around, until now. Instead of hard coding software routines with specific instructions to accomplish a particular task, machine learning is a way of “training” an algorithm so that it can learn how.
The world’s first deep learning enabled video camera is here, and it’s called AWS DeepLens. The DeepLens is a deep learning camera from Amazon Web Services (AWS) that helps put deep learning in the hands of inexperienced and experienced developers. It’s a fully programmable video camera that comes with tutorials, code and pre-trained machine learning models, all design to expand ones’ deep learning skills.
Once you have your AWS account set up it only takes 10 minutes to register the device and launch your first project. Either you want to customize your own model or use one of the pre-trained ones. It comes with 7 sample projects e.g. face detection, object recognition or activity recognition and more. As an unexperienced deep learning developer, I found the DeepLens easy to use and a door opener to the world of deep learning.
This is a screenshot from the live stream of the face detection model from the DeepLens. It detects all faces in the camera stream direct from the DeepLens and displays a percentage of it being a face. Further development with this pre-trained model could be to enable it to recognize the faces instead of only detecting them.
And this is a screenshot of the object recognition project. As you can see in the picture it captures 1/3 tv monitors and a chair, the percentages are not high, but enough for the camera to consider it of being objects.
The screenshots above is two of the sample projects that comes with the camera. The DeepLens can be combined with other AWS services such like, the AWS Sagemaker to launch your trained models and then view them on the DeepLens. This opens a door to customize your own projects, everything from teaching it how to read a book to games to categorize vehicle types.
With this new technology everyone can get into deep learning, and as your knowledge base grows it could evolve into more advanced deep learning models and with that, new possibilities.