Hi, I'm Ankit currently I enrolled for AWS Training and Certification.
Transcript
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
Hi. I'm Scott Gilbert
with AWS Training and Certification.
Welcome to Amazon Rekognition.
In this video, we will discuss Rekognition,
part of Amazon's machine learning offerings.
I will highlight how easy it is to get started with
Rekognition as well as give
an introduction to its programmatic use.
I've been with AWS for just under two years and I'm
currently responsible for delivering
training across North America.
As part of the training team,
I've contributed to delivering classes as well
as assisting in the creation of new content.
Rekognition is one of
AWS's offerings in artificial intelligence.
Rekognition allows you to easily analyze images.
Your applications can make calls to
Rekognition allowing you to recognize faces,
landscapes, and even the mood of a person.
In this video, I will introduce
to you the service Rekognition.
I will talk to you about the benefits of the service as
well as go through the key features of Rekognition.
I will step through a couple of use cases of Rekognition,
highlighting just how powerful the service can be.
I will end this presentation with
a quick demonstration of the service,
and I hope to have you logging into
AWS to use this as well.
The main benefits of Rekognition are it's
ease of use as well as its low cost.
With zero experienced in developing machine learning,
you can load an image into
Rekognition and it will be analyzed automatically.
You can take advantage of the APIs
we've created to automate this analysis.
Best of all, as a managed service,
Amazon will handle auto-scaling
of Rekognition allowing you to
potentially send thousands of images
an hour for analysis through recognition.
The key features of Rekognition
are object detection where
the engine is able to accurately
determine what isn't an image,
facial analysis, facial comparison,
as well as facial recognition.
The Engine is able to determine mood,
the number of faces,
as well as differentiating
different faces from each other.
All processed images give you a confidence score,
giving you the ability to drop low confidence processes
or perhaps flag them for secondary or even manual review.
Let's look at a few use cases for recognition.
Here, I hope to showcase just how useful
and powerful the service can be for your applications.
Searchable image libraries can be useful for
many cases such as
real estate agent looking to
take a picture of properties.
They can then upload these images to an S3 bucket.
This upload automatically triggers a lambda function.
The Lambda function calls Rekognition
which pulls the image from the S3 bucket,
and analyze this image and returns the image and
labels along with the confidence scores for
each label into Elastic Search.
Users can now search
through the labels to find the images.
This makes searching for keywords such as a garage or
yard far far easier for any real estate application.
Another use case would be image moderation.
This functionality can greatly assist foreign moderators
ensuring that inappropriate content is not posted.
As a user uploads an image,
the object upload triggers a lambda function.
Lambda calls Rekognition which
examines the image for the content.
If there is no inappropriate content found,
the picture is immediately posted and viewable by users.
If there is inappropriate content detected,
the post is either rejected and
the image is removed from storage.
If the confidence scores are not high enough,
either way it can also send the image for
manual review ensuring that
only approved content is shown.
Sentiment analysis can be quite useful for many users.
From retailers wanting to track the effectiveness of
sales or demographic information of the shoppers,
to a company that just wants to
ensure that their workforce is happy.
Live images are captured via
On-premise camera or user uploaded image.
The application will send
the images to Rekognition for our analysis.
The app stores the information in S3 and
Redshift will allow them to receive this data in.
Quick site can then be used for
regular analysis of the information,
tracking the sentiment over time.
All information can then be placed into
a weekly or monthly or even quarterly marketing report
to track the effectiveness of your ongoing sales,
upgrades to the stores,
or just changes in your overall merchandise over time.
I'm now going to step through
the Rekognition engine on AWS.
I hope to highlight just how easy this is to use.
So I've logged into the management console.
Let's go ahead and load up the Rekognition service.
If you've recently visited the Engine,
you will see it here in your recently visited Services.
Otherwise, you can click on
services and open up the Rekognition Engine.
You can access Rekognition programmatically and
use it's APIs to analyze your images.
You can download the SDK and run
your programs or write programs that utilize Rekognition.
I'm just going to try the demo to Rekognition to
highlight just how easy it is to analyze your images.
As we pull up Rekognition,
we have some stock images
that are already loaded into this.
On the right, it will show the results.
You have the labels as well as
the confidence score for each label.
There is, in fact, a skateboard,
there's people in this image,
and there's cars parked on the side of the road.
We want to expand this a little bit more.
We see automobiles, vehicle, intersection.
It is in fact on a road all the way
down to it's in a metropolis,
there's SUVs and apartments.
This is a stock image.
Let's go ahead and try to upload our own.
Click on the "Upload" button and select an image.
Now, this is an image of Seattle.
Our labels found are architecture, city,
downtown, high rise, skyscraper,
as well as Metropolis.
We have confidence scores for all of these labels.
This is, in fact, a architecture is shown in a city.
It is downtown. There are multiple high rises.
It is, in fact, in a Metropolis.
This image is fairly easy.
Let's look at another image.
This is multiple people in here.
We have people, humans, clothing, overcoat.
We want to go through facial analysis.
In this image, we are detecting
a face from the rest of the image.
We want to load in.
It is able to pull and highlight where the faces are.
Let's see if any of these people are celebrities.
Well, I know this person is a celebrity.
Let's go back to our old image.
But unfortunately, none of
the people in this image are celebrities.
So as you can see,
Rekognition is quite easy to use.
There are multiple tools
multiple demos that are available.
I really hope that you log on and try to play
around with different images
and the analysis of the images.
In summary, Rekognition is
a great tool in our machine learning offerings.
Rekognition gives you the ability to
run multiple forms of analysis on
your images and returns labels with
confidence scores to ensure the analysis is accurate.
I hope you've learned a little something and we'll
continue to explore our other videos.
I'm Scott Gilbert, with
AWS Training and Certification. Thank you for watching.
Transcript Help Us Translate Interactive Transcript - Enable basic transcript mode by pressing the escape key You may navigate through the transcript using tab. To save
.jpeg)
0 Comments