What is AWS Deep Lens Process | Full Course | AWS Deep Lens Tutorial for Beginners |

 

This video on "MAWS Deep Lens" will provide you with a comprehensive and detailed knowledge of AWS Deep Lens concepts with hands-on examples.

Transcript

Select a languageArabicEnglishFrenchGermanItalianPortuguese (European)RussianSpanishVietnamese

Interactive Transcript - Enable basic transcript mode by pressing the escape key

You may navigate through the transcript using tab. To save a note for a section of text press CTRL + S. To expand your selection you may use CTRL + arrow key. You may contract your selection using shift + CTRL + arrow key. For screen readers that are incompatible with using arrow keys for shortcuts, you can replace them with the H J K L keys. Some screen readers may require using CTRL in conjunction with the alt key

Welcome to this introductory course on AWS DeepLens. 

The world's first deep learning 

enabled video camera for developers. 

I'm Jyothi Nookula and I'm 

a senior product manager on the DeepLens team. 

Today, I'll show you how to get 

started building AWS DeepLens projects. 

We'll look at the device's hardware, 

the architecture, and some sample project templates. 

At the end, I'll demonstrate building a project using 

the object detection template from 

the AWS DeepLens console. Let's get started. 

At Amazon, we have been working on 

machine learning for 20 years. 

We know there's no easy way for you 

as a developer to learn machine learning, 

and test and prototype your ideas. 

So we decided to put deep learning 

into your hands literally. 

AWS DeepLens is a wireless enabled video camera 

and a development platform that's 

integrated with the AWS Cloud. 

It lets you use the latest 

artificial intelligence tools and 

technology to develop computer vision applications 

based on deep learning models. 

Before we start building projects, 

let's take a look at the actual device. 

What makes our AWS DeepLens stand apart is the on-board 

accelerator that is capable of 

delivering 100 gigaflops of compute, 

which means it can run 100 billion operations per second. 

So how do you take this device 

and actually build something? 

For that, you need to understand 

the general workflow for an AWS DeepLens deployment. 

To create and run an AWS DeepLens project, 

you typically use Amazon SageMaker, 

AWS lambda, and AWS green grass. 

You'd use Amazon SageMaker to train and validate 

a custom machine learning model 

or you could import a pre-trained model. 

AWS lambda functions in DeepLens 

perform three important operations. 

Pre-processing, 

capturing inference, and displaying output. 

Once a project is deployed to DeepLens, 

the model and the lambda function 

can run locally on the device. 

AWS DeepLens creates 

a computer vision application project 

that consists of the model and inference lambda function. 

AWS Green grass can deploy 

the project and a lambda run-time to your AWS DeepLens, 

as well as the software or configuration updates. 

This diagram illustrates how 

these services come together. 

First, when turned on, 

AWS DeepLens captures a video stream, 

it produces two output streams. 

The device stream, which is the video stream 

passed through without processing and the project stream, 

which contains the results of 

the model's processing video frames. 

From there, the inference lambda function 

receives unprocessed video frames and 

then passes those unprocessed frames to 

the project's deep learning model for processing. 

Finally, the inference lambda function receives 

the process frames back from 

the model and passes them on in the project stream. 

We don't really show it in the diagram, 

but AWS DeepLens is well 

integrated with other AWS services. 

For instance, projects deployed to AWS DeepLens 

are securely transferred using AWS Greengrass. 

The output of AWS DeepLens 

when connected to the Internet, 

can be sent back to the console via 

AWS IoT and Amazon Kinesis video streams. 

So now, we're ready to start building, right? 

The reality is, most of us don't have 

the skills to build a convolutional neural network model. 

That's why AWS DeepLens 

offers ready to deploy sample projects, 

which include a pre-trained 

convolutional neural network model 

and the corresponding inference function. 

AWS DeepLens offers 

seven sampled project templates ready-to-use, 

including templates such as object detection, 

hot dog/not hot dog, 

artistic style transfer, 

activity recognition, and face detection. 

With these sample projects, 

you can get started with 

machine learning in less than 10 minutes. 

These templated projects can be edited and extended. 

For instance, you could use 

the object detection project template 

to recognize when your dog is 

sitting on your couch and have the application 

send you an SMS to notify you of this event. 

Of course, you can also create, train, 

and deploy your own custom model to AWS DeepLens. 

The development platform supports a variety of 

deep learning frameworks including 

MXNet, TensorFlow, and Caffe. 

Now, let's take it for a spin. 

We'll deploy the object detection project. 

The project uses 

the single shot multi box detective framework 

to detect objects with a pre-trained resonate 15 network. 

The network has been trained on 

the Pascal VOC dataset and 

can recognize 20 different objects. 

The model takes the video stream from your AWS DeepLens 

as input and labels the objects that it identifies. 

It's a pre-trained optimized model that 

is ready to be deployed to your AWS DeepLens. 

After deploying it, you can review the objects 

your AWS DeepLens recognizes 

around you through the console. 

Let's see what it looks like in action. 

I'm starting here in the AWS management console. 

In the search bar, 

I'm going to type DeepLens.

Now, we're looking at the project screen. 

If you don't see the project list view, 

click on the ''Hamburger menu'' on 

the left and select projects.

Select "Create new project" on the top right, 

choose "User project template" and "Object detection", 

and scroll down the screen to choose next. 

I'm going to accept the default values, 

scroll to the bottom and select "Create". 

It can take a few moments for 

the project creation to fully complete. 

You can verify that the project was created 

successfully once the description field has a value. 

You may need to refresh 

your browser window if this value isn't populating. 

Now that the fields are populated, 

choose the radio button for 

the project and select deploy to device. 

On the target device screen, 

choose your device from the list and select review. 

Now, it's time to review the details of 

the deployment and select "Deploy".

The process to deploy the model to 

your DeepLens can take a few minutes. 

You will know that it is completed 

when the blue banner changes to green. 

In the project output tab, 

copy the unique topic for your AWS DeepLens. 

Choose AWS IoT console and paste the unique topic,

to subscribe to a topic. 

You can review the results of DeepLens detecting 

different objects in the room and 

printing the confidence code tool. 

Notice, as I change what the DeepLens can see, 

it will report that at the bottom of the screen.

Now, let's change it to a bottle.

Now, you have seen the power of DeepLens. 

I'm Jyothi Nookula with AWS. 

Thanks for watching.

Transcript Help Us Translate Interactive Transcript - Enable