AWS DeepLens - A Complete Guide

AWS DeepLens is a Video camera looking device that applies the neural networks to learn and make predictions via computer vision projects, real-world, hands-on exploration with a physical device and tutorials. This device performs like an intelligent device that will operate deep learning algorithms on captured images in real-time.

The difference between the other Artificial Learning powered camera and the Deep Lens is the features and capabilities that don't require sending video frames to the cloud and making it possible to run machine learning inference models locally. This device shapes up the theories and hypotheses on edge computing as to how advanced it can be with further development. 

Now even though if the DeepLens device looks like a video camera but in reality, it works more like a PC and less like a camera. It’s a powerful computer that has an attached camera which is much more advanced compared to the average webcam. 

Specification:

AWS DeepLens is installed with an Intel Atom X5 Processor that comes with four threads and four cores. It comes with 16 GB Storage and 8 GB RAM computing resource which is enough to run the machine learning algorithms. DeepLens is embedded with GPU in the form of Intel Gen9 Graphics Engine which may not be the best hardware but it is enough to run local Machine Learning inferences. 

The device operates Ubuntu 16.04 LTS, AWS Greengrass Core and optimized version of Intel cLDNN libraries and it also can be connected to a mouse, HDMI Display, and standard keyboard. You can activate the terminal window and run the device like any other Linux Machine. The camera is 4-megapixel webcam with a 1080 resolution with a 2D microphone array. 

It is designed for developers to run models with TensorFlow and Caffe in less than 10 minute start-up time. Basically, the device is been build so that to put forth the machine learning concept into the field to make it familiar with the developers and general users. 

Benefits of AWS DeepLens:

Custom built:

With keeping Deep learning in mind the DeepLens is been designed with over 100 GFLOPS of compute power on the device with the ability to process deep learning prediction in real time on HD Video. 

Fully programmable:

AWS DeepLens is fully programmable using the AWS Lambda functions and is easily customizable. The Deep Learning models in the DeepLens run as part of the AWS Lambda functions that deliver a familiar programming environment to experiment with. 

A new approach to learn machine learning:

AWS DeepLens enables developers of all skill levels to work on deep learning with sample projects with practical and hands-on the example that can start running with just a single click. 

Integrated with Amazon SageMaker:

From the AWS Management Console, the models trained in the Amazon SageMaker can be transferred to the AWS DeepLens with just a few clicks. 

Broad Framework support:

AWS Developers can operate any Deep Learning foundation including the TensorFlow and Caffe. AWS DeepLens use Apache MXNet to deliver a high performance, optimized and efficient inference engine. 

AWS integrated platform:

AWS DeepLens is integrated with Amazon Recognition for advanced image analysis; Amazon Polly to build speech-enabled projects and Amazon SageMaker for training models. The device is also connected with Amazon Simple Notification Service, Amazon Simple Storage Service, AWS Internet of Things, Amazon DynamoDB, Amazon SQS and more. 

You can build a lot of skills using AWS DeepLens but as of now, there are collections of projects that are created by the developer community so that you can use it for inspiration to design different or similar models. 

Some of the things that you can do with AWS DeepLens:

• You can accurately recognize and detect objects.
• Classify food whether specifying that either it is a hot dog or not a hot dog. 
• Using DeepLens Identify whether a cat or dog.
• You can transfer the style from one image to an entire video sequence recorded by the DeepLens in real time.
• With DeepLens you can identify more than 30 actions such as playing guitar, dancing, brushing teeth or applying lipstick. 
• You can detect the faces of people.
• You can detect 9 different head movement orientation. 

The above-mentioned skills only are few chosen but there are more collections of DeepLens projects that are created by the developer community. You can check them out and get inspired to build a new skill.

 AWS DeepLens is now available for $249.