Q: What is AWS DeepLens?

AWS DeepLens is the world’s first deep-learning enabled video camera for developers of all skill levels to grow their machine learning skills through hands-on computer vision tutorials, example code, and pre-built models.

Q: How is AWS DeepLens different from other video cameras in the market?

AWS DeepLens is the world's first video camera optimized to run machine learning models and perform inference on the device. It comes with 6 sample projects at launch, that you can deploy to your AWS DeepLens in less than 10 minutes. You can run the sample projects as is, connect them with other AWS services, train a model in Amazon Sagemaker and deploy it to AWS DeepLens, or extend the functionality by triggering a lambda function when an action takes place. You can also apply more advanced analysis on the cloud using Amazon Rekognition. AWS DeepLens provides the building blocks for your machine learning needs.

Q: What sample projects are available at launch?

There are 6 sample projects available at launch. We will continue to launch practical and fun projects for developers to use and learn, based on user feedback. The 6 sample projects are:

1. Object Detection
2. Hot Dog Not Hot Dog
3. Cat and Dog
4. Artistic Style Transfer
5. Activity Detection
6. Face Detection


Q: What geographic regions is AWS DeepLens available?

Currently AWS DeepLens is only available in the US.


Q: Does AWS DeepLens include Alexa?

No, AWS DeepLens does not have Alexa or any far-field audio capabilities. However, AWS DeepLens has a 2D microphone array that is capable of running custom audio models, with additional programming required.


Q: How can I get a AWS DeepLens?

Currently, AWS DeepLens is available to select participants of AI/ML sessions at AWS re:Invent 2017, and is available for pre-order at:


Product Details

Q: What are the product specifications of the device?

  • Intel Atom® Processor
  • Gen9 graphics
  • Ubuntu OS 16.04 LTS
  • 106 GFLOPS performance
  • Dual band Wi-Fi
  • 8GB RAM
  • 16GB memory
  • Expandable storage via microSD card
  • 4MP camera with MJPEG
  • H.264 encoding at 1080p resolution
  • 2 USB ports
  • Micro HDMI
  • Audio out

Q: What deep learning frameworks can I run on the device?

AWS DeepLens is optimized for Apache MXNet. Support for TensorFlow and Caffe will be available in the future.

Q: What kind of performance can I expect with AWS DeepLens?

Performance is measured on images inferred per second and latency. Different models will have varying inference per second. The baseline inference performance is 14 images/second on AlexNet, and 5 images/second on ResNet 50 for batch size of 1. The characteristics of the network that the DeepLens is connected to will determine the latency performance.

Q: What MXNet network architecture layers does AWS DeepLens support?

AWS DeepLens offers support for 20 different network architecture layers. The layers supported are:

  • Activation
  • BatchNorm
  • Concat
  • Convolution
  • elemwise_add
  • Pooling
  • Flatten
  • FullyConnected
  • InputLayer
  • UpSampling
  • Reshape
  • ScaleShift
  • SoftmaxActivation
  • SoftmaxOutput
  • transpose
  • _contrib_MultiBoxPrior
  • _contrib_MultiBoxDetection
  • _Plus
  • Deconvolution
  • _mul

Getting Started

Q: What comes in the box and how do I get started?

Inside the box, developers will find a Getting Started guide, the AWS DeepLens device, a power supply and a 32GB microSD card. Setup and configuration of the DeepLens device can be done in minutes using the AWS DeepLens console, and by configuring the device through a browser on your laptop or PC.

There are four 10-Minute Tutorials designed to help guide you through getting started:

1. Configure Your New AWS DeepLens
2. Create and Deploy a Project
3. Extend a Project
4. Build an AWS DeepLens Project with Amazon SageMaker


Q: Can I train my models on the device?

No, AWS DeepLens is capable of running inference or predictions using trained models. You can train your models in Amazon SageMaker, a machine learning platform to train and host your models. AWS DeepLens offers a simple 1-click deploy feature to publish trained models from Amazon SageMaker.

Q: What AWS services are integrated with AWS DeepLens?

DeepLens is pre-configured for integration with AWS Greengrass, Amazon SageMaker and Amazon Kinesis Video Streams. You can integrate with many other AWS services, such as Amazon S3, Amazon Lambda, Amazon Dynamo, Amazon Rekognition using AWS DeepLens.

Q: Can I SSH into AWS DeepLens?

Yes, we have designed AWS DeepLens to be easy to use, yet accessible for advanced developers. You can SSH into the device using the command: ssh aws_cam@


Q: What programming languages are supported by AWS DeepLens?

You can define and run models on the camera data stream locally in Python 2.7.

Q: Do I need to be connected to internet to run the models?

No. You can run the models that you have deployed to AWS DeepLens without being connected to the internet. However, you need internet to deploy the model from the cloud to the device initially. After transferring your model, AWS DeepLens can perform inference on the device locally without requiring cloud connectivity. However, if you have components in your project that require interaction with cloud, you will need to have internet for those components.

Q: Can I run my own custom models on AWS DeepLens?

Yes. You can also create your own project from scratch, using the AWS SageMaker platform to prepare data and train a model using a hosted notebook, and then publish the trained model to your AWS DeepLens for testing and refinement. You can also import an externally-trained model into AWS DeepLens by specifying the S3 location for model architecture and network weights files.