Difference between revisions of "DeepLens - deep learning enabled video camera"
| Line 62: | Line 62: | ||
<youtube>5DDRvRHZ1Qs</youtube> | <youtube>5DDRvRHZ1Qs</youtube> | ||
<youtube>wnTvVB1ojPk</youtube> | <youtube>wnTvVB1ojPk</youtube> | ||
| − | |||
| − | |||
<youtube>tu7TmUQoj6s</youtube> | <youtube>tu7TmUQoj6s</youtube> | ||
| Line 69: | Line 67: | ||
* [http://aws.amazon.com/blogs/machine-learning/deploy-gluon-models-to-aws-deeplens-using-a-simple-python-api/ Deploy Gluon models to AWS DeepLens using a simple Python API] | * [http://aws.amazon.com/blogs/machine-learning/deploy-gluon-models-to-aws-deeplens-using-a-simple-python-api/ Deploy Gluon models to AWS DeepLens using a simple Python API] | ||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
== I just got my DeepLens! == | == I just got my DeepLens! == | ||
| Line 90: | Line 81: | ||
<youtube>nINqpklf7Eo</youtube> | <youtube>nINqpklf7Eo</youtube> | ||
| + | == First Project == | ||
| + | |||
| + | [http://docs.aws.amazon.com/deeplens/latest/dg/deeplens-templated-projects-overview.html This project] shows you how a deep learning model can detect and recognize objects in a room. The project uses the Single Shot MultiBox Detector (SSD) framework (Reference: [[Object Detection; Faster R-CNN, YOLO, SSD]]) to detect objects with a pretrained resnet_50 network. The network has been trained on the Pascal VOC dataset and is capable of recognizing 20 different kinds of objects. The model takes the video stream from your AWS DeepLens device as input and labels the objects that it identifies. The project uses a pretrained optimized model that is ready to be deployed to your AWS DeepLens device. After deploying it, you can watch your AWS DeepLens model recognize objects around you. The model is able to recognize the following objects: airplane, bicycle, bird, boat, bottle, bus, car, cat, chair, cow, dining table, dog, horse, motorbike, person, potted plant, sheep, sofa, train, and TV monitor. | ||
| + | |||
| + | <youtube>5VAKcQtoELo</youtube> | ||
| + | <youtube>SZTo7CqOjvU</youtube> | ||
== SSH == | == SSH == | ||
| Line 97: | Line 94: | ||
<youtube>2eKjcLsBH6E</youtube> | <youtube>2eKjcLsBH6E</youtube> | ||
<youtube>HozP1t3usPM</youtube> | <youtube>HozP1t3usPM</youtube> | ||
| + | |||
| + | == Processor == | ||
| + | |||
| + | With over 100 GFLOPS of compute power on the device, it can process deep learning predictions on HD video for real time. | ||
| + | |||
| + | * [http://software.intel.com/sites/default/files/managed/c5/9a/The-Compute-Architecture-of-Intel-Processor-Graphics-Gen9-v1d0.pdf Processor Technical Specifications] | ||
| + | <youtube>suQnh1TvGHw</youtube> | ||
| + | <youtube>TS4ShpBHr_g</youtube> | ||
Revision as of 07:29, 24 June 2018
- Product page...
- AWS DeepLens Workshop: Building Computer Vision Applications
- AWS DeepLens meets SORACOM; cellular enabled cloud-connectivity-as-a-service platform
Contents
Integrated Components/Technologies
- Lex - conversational interfaces using voice and text
- SageMaker - build, train, and deploy
- Polly - text to speech
- Rekognition - video analysis service
- Kinesis - collect, process, and analyze real-time, streaming data
- Lambda - run code without managing servers
- AWS IoT Services Overview
- Internet of Things (IoT) Core - process and route those messages to AWS endpoints
- AWS IoT Button
- Greengrass - connected devices can run AWS Lambda functions, keep device data in sync
- Intel® Compute Library for Deep Neural Networks (clDNN) & OpenVINO - deep learning primitives for computer vision
- Simple Queue Service (SQS) - message queuing
- Simple Notification Service (SNS) - pub/sub messaging and mobile notifications
- DynamoDB - NoSQL database
- Simple Storage Service (S3) - object storage
- Management Console - manage web services
- Deep Learning (DL) Amazon Machine Image (AMI) - DLAMI
- SoftAP - software enabled access point
- Ubuntu - operating system
Frameworks
Projects
- aws-ml-vision-end2end - Jupyter Notebook tutorials walking through deep learning Frameworks (MXNet, Gluon) to Platforms (SageMaker, DeepLens) for common CV use-cases | GitHub
- Deploy Gluon models to AWS DeepLens using a simple Python API
I just got my DeepLens!
- DeepLens Developer Resources
- Part 1 - Getting Online | SAMZAKKOUR @ Kloud
- Lynda.com Setting up your AWS DeepLens
First Project
This project shows you how a deep learning model can detect and recognize objects in a room. The project uses the Single Shot MultiBox Detector (SSD) framework (Reference: Object Detection; Faster R-CNN, YOLO, SSD) to detect objects with a pretrained resnet_50 network. The network has been trained on the Pascal VOC dataset and is capable of recognizing 20 different kinds of objects. The model takes the video stream from your AWS DeepLens device as input and labels the objects that it identifies. The project uses a pretrained optimized model that is ready to be deployed to your AWS DeepLens device. After deploying it, you can watch your AWS DeepLens model recognize objects around you. The model is able to recognize the following objects: airplane, bicycle, bird, boat, bottle, bus, car, cat, chair, cow, dining table, dog, horse, motorbike, person, potted plant, sheep, sofa, train, and TV monitor.
SSH
Processor
With over 100 GFLOPS of compute power on the device, it can process deep learning predictions on HD video for real time.
