DeepLens is essentially a small computer based on Ubuntu and Intel Atom with a built-in camera that is easy enough to run and evaluate visual machine learning models. DeepLens has a computational performance of approximately 106 GFLOPS.
This hardware has all the usual I/O ports (such as Micro HDMI, USB 2.0, audio output, etc.), allowing you to create prototype applications, whether these applications are simple toy applications or complex industrial applications. One of the 4 megapixel cameras is sufficient for most use cases. Not surprisingly, DeepLens is deeply integrated with other AWS services. This includes Greengrass, the AWS IoT service for deploying models to DeepLens, and SageMaker, which isAmazonThe latest tool for building machine learning models.
These integrations also make it easy to get started with the camera. In fact, if you only want to run one of the prebuilt examples provided by AWS, you should not set DeepLens more than 10 minutes and deploy one of the models to the camera. These project templates include an object detection model that can distinguish 20 objects, a style conversion example, rendering camera images in Van Gogh style, a face detection model and model that can recognize cats and dogs, and one that can identify approximately 30 species Different action models. The DeepLens team also added models for tracking head poses.
DeepLens team emphasizes that even if you have never used machine learningDevelopmentPeople can also use existing templates and easily extend them. This is partly due to the fact that the DeepLens project consists of two parts: the model and the Lambda function that runs the model instance, and allows the user to perform operations based on the model's output. With SageMaker, AWS now offers a tool that makes it easy to build models without having to manage the underlying infrastructure.