AWS DeepLens is a programmable video camera that enables a developer to easily experiment with machine learning, artificial intelligence (AI) and the Internet of Things (IoT). The DeepLens camera integrates with AI services hosted on the AWS public cloud.
AWS DeepLens comes with multiple tutorials and sample projects to get started with a subset of machine learning technology called deep learning. A developer with limited deep learning or AI experience can use these programs to understand the basics of the technology. A developer who is more familiar with deep learning can use the device to deploy an application with audio and visual recognition capabilities.
DeepLens supports the TensorFlow, Caffe and Apache MXNet deep learning frameworks.
AWS DeepLens physical specs
The DeepLens device stands at just under 7 inches tall and includes a 4-megapixel, 1080p camera. It has 8 GB of on-board memory and 16 GB of storage capacity, and comes with a 32 GB card for additional storage capabilities.
The three sensors on the front of the DeepLens base indicate camera status, network/Wi-Fi status and overall power status. The power on/off button is also located on the front of the base.
The back of the DeepLens base features two USB 2.0 ports, a micro SD card slot, a micro HDMI slot, a reset button, an audio outlet and a power supply connector.
AWS DeepLens sample projects
Out of the box, AWS DeepLens comes with eight sample deep learning projects. These projects are designed to teach the user the basics of DeepLens, and to cover some of the most popular computer vision use cases. These sample projects include:
- Object Detection - to detect and identify different objects;
- Hot Dog Not Hot Dog - to classify food as a hot dog or not a hot dog;
- Cat and Dog - to identify a dog or a cat;
- Artistic Style Transfer - to transfer one image's style to a video sequence in real time;
- Activity Recognition - to recognize up to 30 different pre-programmed physical activities;
- Face Detection - to detect faces of people in pictures or video;
- Head Pose Detection - to identify up to nine different pre-programmed head pose angles;
- Community Projects – to use other projects created by the developer community.
AWS DeepLens integration capabilities
DeepLens integrates with multiple AWS offerings to enable further experimentation with deep learning technologies. For example, a developer can build custom models in Amazon SageMaker and send them to DeepLens through the AWS Management Console.
DeepLens also works with Amazon Rekognition for image analysis and Amazon Polly to create speech-enabled projects. Other integrations include Amazon Simple Notification Service, Amazon Simple Queue Service, AWS IoT, Amazon S3, AWS Lambda and DynamoDB.