It’s the season for shopping and AWS DeepLens caught my attention this morning. AWS DeepLens is a deep-learning enabled developer kit with a video camera. The video camera is fully programmable and designed to expand deep learning skills.
DeepLens has a 4MP camera powered by Intel Atom Processor with more than 100 gigaflops of power. It can capture 1080P video along with a 2D microphone array. The DeepLens has WiFi, USB, micro HDMI ports and 8 gigabytes of memory.
The DeepLens gives a great arena to learn deep learning and IoT as well as build cool apps for business uses. You can use AWS Greengrass, AWS Lambda and other AWS AI and infrastructure for building a sophisticated app to solve real world problems.
Developers can use hands-on tutorials, pre-built models to learn and can extend them to other scenarios. It comes loaded with cool project templates such as Face detection, Object detection such as TV, bottles; Action recognition such as brushing teeth, applying lipstick, playing basketball or guitar; Cat & dog recognition, Hot dog recognition and Artistic style transfer.
The DeepLens is an edge device as the incoming video can be used along with the cloud for more higher-level processing. The face detection on DeepLens can be combined with Amazon Rekognition to give a powerful real world solution. You can customize and display the DeepLens output to a laptop, extend it to send SMS notifications with AWS Lambda, stream video back to AWS using Amazon Kinesis Video Streams, and apply more advanced video analytics using Amazon Rekognition Video. The device also connects securely to AWS IoT, Amazon SQS, Amazon SNS, Amazon S3, Amazon DynamoDB, and more. You can run any deep learning framework, including TensorFlow and Caffe. AWS DeepLens comes pre-installed with a high performance, efficient, optimized inference engine for deep learning using Apache MXNet.
Now coming to the application of this technology, it can help you find the number of people walking in front of a store or watching your advertisement display, celebrity recognition in the crowd, search all objects and faces detected within a timeline or frame, etc
Oops! Burnt my breakfast again and today the reason is DeepLens. I could certainly benefit from a DeepLens with a DeepSmell which can inform me that my dish is about to burn. Can AWS SMELL something new in AI here?