Media Insights on AWS is a development framework for building serverless applications that process video, images, audio, and text on AWS. It takes care of workflow orchestration and data persistence so that you can focus on workflow development. By addressing the concerns of running workflows, Media Insights on AWS empowers you to build applications faster with the benefit of inheriting a pre-built and robust back end.
Media Insights on AWS has been successfully used in a variety of scenarios, such as:
- Deriving video features for ad placement
- Transforming video content with redaction
- Indexing videos based on visual and audio content
- Translating videos for automated localization
For additional details and sample use cases, refer to How to Rapidly Prototype Multimedia Applications on AWS with the Media Insights on AWS on the AWS Media blog.
This repository contains the Media Insights on AWS back-end framework. Users interact with the framework through REST APIs or by invoking Lambda functions directly. You will not find a graphical user interface (GUI) in this repository, but a reference application for Media Insights on AWS that includes a GUI is in the Content Localization repository.
You can deploy Media Insights on AWS in your AWS account with the following Cloud Formation templates. The Cloud Formation stack name must be 12 or fewer characters long.
| Region | Launch |
|---|---|
| US East (N. Virginia) |  |
| US West (Oregon) | |
| EU West (Ireland) | |
The Cloud Formation options for these one-click deploys are described in the installation parameters section.
Run the following commands to build and deploy Media Insights on AWS from scratch. Be sure to define values for MI_STACK_NAME and REGION first.
REGION=[specify a region]
MI_STACK_NAME=[specify a stack name]
git clone https://github.com/aws-solutions/media-insights-on-aws
cd media-insights-on-aws
cd deployment
VERSION=0.0.0
DATETIME=$(date '+%s')
DIST_OUTPUT_BUCKET=media-insights-on-aws-$DATETIME
aws s3 mb s3://$DIST_OUTPUT_BUCKET-$REGION --region $REGION
aws s3 mb s3://$DIST_OUTPUT_BUCKET --region $REGION
./build-s3-dist.sh --template-bucket $DIST_OUTPUT_BUCKET --code-bucket $DIST_OUTPUT_BUCKET --version $VERSION --region $REGION
TEMPLATE={copy "Template to deploy" link from output of build script}
aws cloudformation create-stack --stack-name $MI_STACK_NAME --template-url $TEMPLATE --region $REGION --capabilities CAPABILITY_IAM CAPABILITY_NAMED_IAM CAPABILITY_AUTO_EXPAND --disable-rollback
If you're building applications on Media Insights on AWS then you will need to understand the following resources in the Outputs tab of the Cloud Formation stack:
- DataplaneApiEndpoint is the endpoint for accessing dataplane APIs to create, update, delete and retrieve media assets
- DataplaneBucket is the S3 bucket used to store derived media (derived assets) and raw analysis metadata created by the workflows.
- WorkflowApiEndpoint is the endpoint for accessing the Workflow APIs to create, update, delete and execute the workflows.
- WorkflowCustomResourceArn is the custom resource that can be used to create workflows in CloudFormation scripts
You are responsible for the cost of the AWS services used while running this solution. The cost for running this solution with the default settings in the us-east-1 (N. Virginia) region is approximately $24 per month without free tiers, or $13 per month with free tiers for 100 workflow runs. Most use cases are covered by the free tier for all AWS services except Amazon Kinesis and AWS Lambda. The costs for the Amazon Kinesis data stream ($12.56/mo) and the Workflow Scheduler lambda ($3.73/mo) will remain relatively unchanged, regardless of how many workflows execute.
| AWS Service | Quantity | Cost |
|---|---|---|
| Amazon API Gateway | 1 million workflows | $3.50 / mo |
| Amazon Dynamo DB | 1 million workflows | $.025 / mo |
| AWS Lambda | 100 workflows | $4.75 / mo |
| Amazon Kinesis | 100 workflows | $12.56 / mo |
| Amazon SQS | 1 million workflows | $0.40 / mo |
| Amazon SNS | n/a | No charge |
| Amazon S3 | 100 workflows | $2.3 / mo |
| AWS Xray | 100 workflows | $.0005 / mo |
These cost estimates are based on workflows processing live action videos 10 minutes in duration. Each additional 100 workflow executions will cost roughly $2, or higher for videos longer than 10 minutes and lower for videos shorter than 10 minutes.
The Cloud Formation stack name for Media Insights on AWS must be 12 or fewer characters long. This will ensure all the resources in the stack remain under the maximum length allowed by Cloud Formation.
Media Insights on AWS does not inherently limit media attributes such as file size or video duration. Those limitations depend on the services used in user-defined workflows. For example, if a workflow uses Amazon Rekognition, then that workflow will be subject to the limitations listed in the guidelines and quotas for Amazon Rekognition. For those who use the Amazon Rekognition service within workflows, be aware about use cases that involve public safety and the general AWS Service Terms.
Deploying Media Insights on AWS builds the following environment in the AWS Cloud:
The AWS CloudFormation template provisions the following resources:
-
Resource: An Amazon API Gateway resource for the control plane REST API
Execution flow: This is the entry point where requests to create, read, update, delete (CRUD), or execute workflows begin.
-
Resource: AWS Lambda and Amazon Simple Queue Service (Amazon SQS) resources to support workflow orchestration and translating user-defined workflows into AWS Step Functions
Execution flow: Requests for workflow CRUD will finish in this step after an AWS Lambda function updates workflow related tables in DynamoDB. Requests to execute workflows will begin in this step by an AWS Lambda function that saves the request to an SQS queue, which is later read and executed by an AWS Lambda function (called, the workflow scheduler) that controls how many workflows can run at the same time.
-
Resource: Amazon DynamoDB tables to store workflow-related data, such as state machine definitions for operators, workflow configurations, and workflow execution status.
-
Resource: Step function resources in AWS Step Functions
Execution flow: When a user defines a new workflow using the workflow API, then an AWS Lambda function creates an executable step function resource in AWS Step Function. When the workflow scheduler starts a workflow, it starts that step function resource, which then invokes a series of AWS Lambda functions that call external services and/or download results from those services. When all the AWS Lambda functions in a workflow have finished execution, then an AWS Lambda function is called to update the workflow status in Amazon DynamoDB.
-
Resource: AWS Lambda functions for using the following commonly used services in workflows: Amazon Rekognition, Amazon Comprehend, Amazon Translate, Amazon Transcribe, Amazon Polly, and AWS Elemental MediaConvert
Execution flow: Operators consist of AWS Lambda functions that call external services and/or download results from those services. They are invoked by a state machine in AWS Step Functions, as prescribed by the workflow definition. These AWS Lambda functions save results to long-term storage via the data plane REST API.
-
Resource: An Amazon API Gateway resource for the data plane REST API
Execution flow: Operators save results to long-term storage by calling this API.
-
Resource: Amazon Simple Storage Service (Amazon S3), DynamoDB, and DynamoDB Streams for media and metadata data storage
Execution flow: The AWS Lambda function behind the data plane API directly accesses Amazon S3 and Amazon DynamoDB to perform incoming CRUD requests. That AWS Lambda function saves files, such as binary media files or JSON metadata files, in Amazon S3. A pointer to those files is saved in an Amazon DynamoDB table. Finally, a time-ordered sequence of modifications to that table are saved in an Amazon DynamoDB Stream and an Amazon Kinesis Data stream.
-
Resource: An Amazon Kinesis Data stream for interfacing with external applications
Execution flow: The Amazon Kinesis Data Streams provides an interface for external applications to access data stored in the data plane. This interface is appropriate for feeding downstream data stores, such as the Amazon Elasticsearch Service or Amazon Neptune, that support specialized data access patterns required by end-user applications. In order to feed a downstream data store, you must implement a consumer (e.g. an AWS Lambda function) that consumes records from the data stream and performs the necessary extract, transform, and load (ETL) tasks needed for the external application.
NOTE: The ETL tasks that feed downstream data stores are entirely use-case dependent and therefore must be user-defined. The Implementation Guide includes detailed instructions for implementing ETL functions in Media Insights on AWS.
-
Workflow API: Use the workflow API to create, update, delete, execute, and monitor workflows.
-
Control plane: The control plane includes the workflow API and state machines for workflows. Workflow state machines are composed of operators from the Media Insights on AWS operator library. When operators within the state machine are run, they interact with the Media Insights on AWS data plane to store and retrieve derived asset and metadata generated from the workflow.
The control plane uses the following Amazon DynamoDB tables store workflow-related data:
Workflow– This table records user-defined workflows.Workflow Execution– This table records the details of every workflow run.Operations– This table records details for each operator in the operator library, such as references to Lambda functions and default runtime parameters.Stage– This table records the auto-generated AWS Step Functions code needed for each operator.System– This table records system-wide configurations, such as maximum concurrent workflows.
-
Operators: Operators are generated state machines that call AWS Lambda functions to perform media analysis or media transformation tasks. Users can define custom operators, but the Media Insights on AWS operator library includes the following pre-built operators:
Celebrity Recognition- An asynchronous operator to identify celebrities in a video using Amazon Rekognition.Content Moderation- An asynchronous operator to identify unsafe content in videos using Amazon Rekognition.Face Detection- An asynchronous operator to identify faces in videos using Amazon Rekognition.Face Search- An asynchronous operator to identify faces from a custom face collection in videos using Amazon Rekognition.Label Detection- An asynchronous operator to identify objects in a video using Amazon Rekognition.Person Tracking- An asynchronous operator to identify people in a video using Amazon Rekognition.Shot Detection- An asynchronous operator to identify camera shots in a video using Amazon Rekognition.Text Detection– An asynchronous operator to identify text in a video using Amazon Rekognition.Technical Cue Detection– An asynchronous operator to identify technical cues such as end credits, color bars, and black bars in a video using Amazon Rekognition.Comprehend Key Phrases– An asynchronous operator to find key phrases in text using Amazon Comprehend.Comprehend Entities– An asynchronous operator to find references to real-world objects, dates, and quantities in text using Amazon Comprehend.Create SRT Captions– A synchronous operator to generate SRT formatted caption files from a video transcript generated by Amazon Transcribe.Create VTT Captions- A synchronous operator to generate VTT formatted caption files from a video transcript generated by Amazon Transcribe.Media Convert- An asynchronous operator to transcode input video into mpeg4 format using AWS Elemental MediaConvert.Media Info– A synchronous operator to read technical tag data for video files.Polly- An asynchronous operator that turns input text into speech using Amazon Polly.Thumbnail- An asynchronous operator that generates thumbnail images for an input video file using AWS Elemental MediaConvert.Transcribe- An asynchronous operator to convert input audio to text using Amazon Transcribe.Translate- An asynchronous operator to translate input text using Amazon Translate.
-
Data plane: This stores the media assets and metadata generated by workflows. Implement a consumer of the Kinesis data stream in the data plane to extract, transform, and load (ETL) data from the master data store to downstream databases that support the data access patterns required by end-user applications.
-
Data plane API: This API is used to create, update, delete, and retrieve media assets and metadata.
-
Data plane pipeline: This pipeline stores metadata for an asset that can be retrieved using an object's
AssetIdandMetadatatype. Writing data to the pipeline initiates a copy of the data to be stored in Kinesis Data Streams. This data stream is the interface that end-user applications can connect to use data stored in the data plane. -
Data pipeline consumers: Changes to the data plane DynamoDB table are reflected in an Amazon Kinesis data stream. For each record in that stream, data pipeline consumers perform the necessary extract, transform, and load (ETL) tasks needed to replicate data, such as media metadata, to the data stores used by external applications. These ETL tasks are entirely use-case dependent and therefore must be user-defined. The Implementation Guide includes detailed instructions for implementing data pipeline consumers.
You can deploy Media Insights on AWS in your AWS account with the one-click deploy buttons shown above.
Stack Name: The name of the stack. This must be 12 or fewer characters long.
| Parameter | Default | Description |
|---|---|---|
MaxConcurrentWorkflows |
5 |
Identifies the maximum number of workflows to run concurrently. When the maximum is reached, additional workflows are added to a wait queue. If too high, then workflows may fail due to external service quotas. Recommended range is 2 to 5. |
DeployAnalyticsPipeline |
true |
Determines whether to deploy a data streaming pipeline that can be consumed by external applications. By default, this capability is activated when the solution is deployed. Set to false to deactivate this capability. |
DeployTestWorkflow |
false |
Determines whether to deploy test resources that contain Lambda functions required for integration and end-to-end testing. By default, this capability is deactivated. Set to true to activate this capability. |
EnableXrayTrace |
false |
Determines whether to activate Active Xray tracing on all entry points to the stack. By default, this capability is deactivated when the solution is deployed. Set to true to activate this capability. |
ExternalBucketArn |
`` | The ARN for Amazon S3 resources that exist outside the stack which may need to be used as inputs to the workflows. The ARN must be a valid Amazon S3 ARN and must reference the same AWS account that is used for the stack. By default, ExternalBucketArn will be blank, meaning workflows will only be able to input media files from the data plane bucket. |
Join our Gitter chat at https://gitter.im/awslabs/aws-media-insights-engine! This public chat forum was created to foster communication between Media Insights on AWS developers worldwide.
For instructions on how to build applications with Media Insights on AWS, read the API reference and builder's guide in the Implementation Guide.
Media Insights on AWS uses AWS_IAM to authorize REST API requests. The following screenshot shows how to test authentication to the Media Insights on AWS API using Postman. Be sure to specify the AccessKey and SecretKey for your own AWS environment.
For more information, see the Implementation Guide.
Amazon Macie can help you discover and protect sensitive data in AWS. If your use-case generates and stores sensitive data to Amazon S3, we recommend that you enable Amazon Macie on the Dataplane Amazon S3 bucket.
To uninstall Media Insights on AWS, delete the CloudFormation stack, as described below. This will delete all the resources created by the Media Insights on AWS template except the Dataplane and the DataplaneLogs S3 buckets. These two buckets are retained when the solution stack is deleted in order to help prevent accidental data loss. You can use either the AWS Management Console or the AWS Command Line Interface (AWS CLI) to empty, then delete those S3 buckets after deleting the CloudFormation stack.
- Sign in to the AWS CloudFormation console.
- Select the Media Insights on AWS stack.
- Choose Delete.
aws cloudformation delete-stack --stack-name <installation-stack-name> --region <aws-region>
Media Insights on AWS creates two S3 buckets that are not automatically deleted. To delete these buckets, use the steps below.
- Sign in to the Amazon S3 console.
- Select the
Dataplanebucket. - Choose Empty.
- Choose Delete.
- Select the
DataplaneLogsbucket. - Choose Empty.
- Choose Delete.
To delete the S3 bucket using AWS CLI, run the following command:
aws s3 rb s3://<bucket-name> --force
This solution collects anonymized operational metrics to help AWS improve the quality and features of the solution. For more information, including how to disable this capability, please see the implementation guide.
It consists of the following information:
- Solution ID: The Media Insights on AWS solution identifier (
SO0163) - Unique ID (UUID): Randomly generated, unique identifier for each Media Insights on AWS deployment
- Timestamp: Date and time of the stack deployment
- Instance Data: The version of the solution that was deployed
Example data:
{
"Solution": "SO0163",
"UUID": "d84a0bd5-7483-494e-8ab1-fdfaa7e97687",
"TimeStamp": "2021-03-01T20:03:05.798545",
"Data": {
"Version": "2.0.5",
"CFTemplate": "Created"
}
}
To opt out of this reporting, complete the following steps before launching the AWS Cloudformation template.
- Download the AWS CloudFormation template to your local hard drive.
- Open the AWS CloudFormation template with a text editor.
- Modify the AWS CloudFormation template mapping section so
SendAnonymizedDatavalue isNo.
Example:
AnonymizedData:
SendAnonymizedData:
Data: No
Visit the Issue page in this repository for known issues and feature requests.
See the CONTRIBUTING file for how to contribute.
The Media Insights on AWS logo features a clapperboard representing multimedia, centered inside a crosshair representing under scrutiny.
See the LICENSE file for our project's licensing.
Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent