Amazon Rekognition Video Analyzer Written in Opencv

Create a Serverless Pipeline for Video Frame Analysis and Alerting

Introduction

Imagine being able to capture live video streams, identify objects using deep learning, and then trigger actions or notifications based on the identified objects -- all with low latency and without a single server to manage.

This is exactly what this project is going to help you accomplish with AWS. You will be able to setup and run a live video capture, analysis, and alerting solution prototype.

The prototype was conceived to address a specific use case, which is alerting based on a live video feed from an IP security camera. At a high level, the solution works as follows. A camera surveils a particular area, streaming video over the network to a video capture client. The client samples video frames and sends them over to AWS, where they are analyzed and stored along with metadata. If certain objects are detected in the analyzed video frames, SMS alerts are sent out. Once a person receives an SMS alert, they will likely want to know what caused it. For that, sampled video frames can be monitored with low latency using a web-based user interface.

Here's the prototype's conceptual architecture:

Architecture

Let's go through the steps necessary to get this prototype up and running. If you are starting from scratch and are not familiar with Python, completing all steps can take a few hours.

Preparing your development environment

Here’s a high-level checklist of what you need to do to setup your development environment.

  1. Sign up for an AWS account if you haven't already and create an Administrator User. The steps are published here.
  2. Ensure that you have Python 2.7+ and Pip on your machine. Instructions for that varies based on your operating system and OS version.
  3. Create a Python virtual environment for the project with Virtualenv. This helps keep project’s python dependencies neatly isolated from your Operating System’s default python installation. Once you’ve created a virtual python environment, activate it before moving on with the following steps.
  4. Use Pip to install AWS CLI. Configure the AWS CLI. It is recommended that the access keys you configure are associated with an IAM User who has full access to the following:
  • Amazon S3
  • Amazon DynamoDB
  • Amazon Kinesis
  • AWS Lambda
  • Amazon CloudWatch and CloudWatch Logs
  • AWS CloudFormation
  • Amazon Rekognition
  • Amazon SNS
  • Amazon API Gateway
  • Creating IAM Roles

The IAM User can be the Administrator User you created in Step 1.

5.   Make sure you choose a region where all of the above services are available. Regions us-east-1 (N. Virginia), us-west-2 (Oregon), and eu-west-1 (Ireland) fulfill this criterion. Visit this page to learn more about service availability in AWS regions.

6.   Use Pip to install Open CV 3 python dependencies and then compile, build, and install Open CV 3 (required by Video Cap clients). You can follow this guide to get Open CV 3 up and running on OS X Sierra with Python 2.7. There's another guide for Open CV 3 and Python 3.5 on OS X Sierra. Other guides exist as well for Windows and Raspberry Pi.

7.   Use Pip to install Boto3. Boto is the Amazon Web Services (AWS) SDK for Python, which allows Python developers to write software that makes use of Amazon services like S3 and EC2. Boto provides an easy to use, object-oriented API as well as low-level direct access to AWS services.

8.   Use Pip to install Pynt. Pynt enables you to write project build scripts in Python.

9.   Clone this GitHub repository. Choose a directory path for your project that does not contain spaces (I'll refer to the full path to this directory as <path-to-project-dir>).

10.   Use Pip to install pytz. Pytz is needed for timezone calculations. Use the following commands:

pip install pytz # Install pytz in your virtual python env

pip install pytz -t <path-to-project-dir>/lambda/imageprocessor/ # Install pytz to be packaged and deployed with the Image Processor lambda function

Finally, obtain an IP camera. If you don’t have an IP camera, you can use your smartphone with an IP camera app. This is useful in case you want to test things out before investing in an IP camera. Also, you can simply use your laptop’s built-in camera or a connected USB camera. If you use an IP camera, make sure your camera is connected to the same Local Area Network as the Video Capture client.

Configuring the project

In this section, I list every configuration file, parameters within it, and parameter default values. The build commands detailed later extract the majority of their parameters from these configuration files. Also, the prototype's two AWS Lambda functions - Image Processor and Frame Fetcher - extract parameters at runtime from imageprocessor-params.json and framefetcher-params.json respectively.

NOTE: Do not remove any of the attributes already specified in these files.

NOTE: You must set the value of any parameter that has the tag NO-DEFAULT

config/global-params.json

Specifies “global” build configuration parameters. It is read by multiple build scripts.

{
    "StackName" : "video-analyzer-stack"
}

Parameters:

  • StackName - The name of the stack to be created in your AWS account.

config/cfn-params.json

Specifies and overrides default values of AWS CloudFormation parameters defined in the template (located at aws-infra/aws-infra-cfn.yaml). This file is read by a number of build scripts, including createstack, deploylambda, and webui.

{
    "SourceS3BucketParameter" : "<NO-DEFAULT>",
    "ImageProcessorSourceS3KeyParameter" : "src/lambda_imageprocessor.zip",
    "FrameFetcherSourceS3KeyParameter" : "src/lambda_framefetcher.zip",

    "FrameS3BucketNameParameter" : "<NO-DEFAULT>",

    "FrameFetcherApiResourcePathPart" : "enrichedframe",
    "ApiGatewayRestApiNameParameter" : "VidAnalyzerRestApi",
    "ApiGatewayStageNameParameter": "development",
    "ApiGatewayUsagePlanNameParameter" : "development-plan"
}

Parameters:

SourceS3BucketParameter - The Amazon S3 bucket to which your AWS Lambda function packages (.zip files) will be deployed. If a bucket with such a name does not exist, the deploylambda build command will create it for you with appropriate permissions. AWS CloudFormation will access this bucket to retrieve the .zip files for Image Processor and Frame Fetcher AWS Lambda functions.

ImageProcessorSourceS3KeyParameter - The Amazon S3 key under which the Image Processor function .zip file will be stored.

FrameFetcherSourceS3KeyParameter - The Amazon S3 key under which the Frame Fetcher function .zip file will be stored.

FrameS3BucketNameParameter - The Amazon S3 bucket that will be used for storing video frame images. There must not be an existing S3 bucket with the same name.

FrameFetcherApiResourcePathPart - The name of the Frame Fetcher API resource path part in the API Gateway URL.

ApiGatewayRestApiNameParameter - The name of the API Gateway REST API to be created by AWS CloudFormation.

ApiGatewayStageNameParameter - The name of the API Gateway stage to be created by AWS CloudFormation.

ApiGatewayUsagePlanNameParameter - The name of the API Gateway usage plan to be created by AWS CloudFormation.

config/imageprocessor-params.json

Specifies configuration parameters to be used at run-time by the Image Processor lambda function. This file is packaged along with the Image Processor lambda function code in a single .zip file using the packagelambda build script.

{
    "s3_bucket" : "<NO-DEFAULT>",
    "s3_key_frames_root" : "frames/",

    "ddb_table" : "EnrichedFrame",

    "rekog_max_labels" : 123,
    "rekog_min_conf" : 50.0,

    "label_watch_list" : ["Human", "Pet", "Bag", "Toy"],
    "label_watch_min_conf" : 90.0,
    "label_watch_phone_num" : "",
    "label_watch_sns_topic_arn" : "",
    "timezone" : "US/Eastern"
}

s3_bucket - The Amazon S3 bucket in which Image Processor will store captured video frame images. The value specified here must match the value specified for the FrameS3BucketNameParameter parameter in the cfn-params.json file.

s3_key_frames_root - The Amazon S3 key prefix that will be prepended to the keys of all stored video frame images.

ddb_table - The Amazon DynamoDB table in which Image Processor will store video frame metadata. The default value,EnrichedFrame, matches the default value of the AWS CloudFormation template parameter DDBTableNameParameter in the aws-infra/aws-infra-cfn.yaml template file.

rekog_max_labels - The maximum number of labels that Amazon Rekognition can return to Image Processor.

rekog_min_conf - The minimum confidence required for a label identified by Amazon Rekognition. Any labels with confidence below this value will not be returned to Image Processor.

label_watch_list - A list of labels for to watch out for. If any of the labels specified in this parameter are returned by Amazon Rekognition, an SMS alert will be sent via Amazon SNS. The label's confidence must exceed label_watch_min_conf.

label_watch_min_conf - The minimum confidence required for a label to trigger a Watch List alert.

label_watch_phone_num - The mobile phone number to which a Watch List SMS alert will be sent. Does not have a default value. You must configure a valid phone number adhering to the E.164 format (e.g. +1404XXXYYYY) for the Watch List feature to become active.

label_watch_sns_topic_arn - The SNS topic ARN to which you want Watch List alert messages to be sent. The alert message contains a notification text in addition to a JSON formatted list of Watch List labels found. This can be used to publish alerts to any SNS subscribers, such as Amazon SQS queues.

timezone - The timezone used to report time and date in SMS alerts. By default, it is "US/Eastern". See this list of country codes, names, continents, capitals, and pytz timezones).

config/framefetcher-params.json

Specifies configuration parameters to be used at run-time by the Frame Fetcher lambda function. This file is packaged along with the Frame Fetcher lambda function code in a single .zip file using the packagelambda build script.

{
    "s3_pre_signed_url_expiry" : 1800,

    "ddb_table" : "EnrichedFrame",
    "ddb_gsi_name" : "processed_year_month-processed_timestamp-index",

    "fetch_horizon_hrs" : 24,
    "fetch_limit" : 3
}

s3_pre_signed_url_expiry - Frame Fetcher returns video frame metadata. Along with the returned metadata, Frame Fetcher generates and returns a pre-signed URL for every video frame. Using a pre-signed URL, a client (such as the Web UI) can securely access the JPEG image associated with a particular frame. By default, the pre-signed URLs expire in 30 minutes.

ddb_table - The Amazon DynamoDB table from which Frame Fetcher will fetch video frame metadata. The default value,EnrichedFrame, matches the default value of the AWS CloudFormation template parameter DDBTableNameParameter in the aws-infra/aws-infra-cfn.yaml template file.

ddb_gsi_name - The name of the Amazon DynamoDB Global Secondary Index that Frame Fetcher will use to query frame metadata. The default value matches the default value of the AWS CloudFormation template parameter DDBGlobalSecondaryIndexNameParameter in the aws-infra/aws-infra-cfn.yaml template file.

fetch_horizon_hrs - Frame Fetcher will exclude any video frames that were ingested prior to the point in the past represented by (time now - fetch_horizon_hrs).

fetch_limit - The maximum number of video frame metadata items that Frame Fetcher will retrieve from Amazon DynamoDB.

Building the prototype

Common interactions with the project have been simplified for you. Using pynt, the following tasks are automated with simple commands:

  • Creating, deleting, and updating the AWS infrastructure stack with AWS CloudFormation
  • Packaging lambda code into .zip files and deploying them into an Amazon S3 bucket
  • Running the video capture client to stream from a built-in laptop webcam or a USB camera
  • Running the video capture client to stream from an IP camera (MJPEG stream)
  • Build a simple web user interface (Web UI)
  • Run a lightweight local HTTP server to serve Web UI for development and demo purposes

For a list of all available tasks, enter the following command in the root directory of this project:

pynt -l

The output represents the list of build commands available to you:

pynt -l output

Build commands are implemented as python scripts in the file build.py. The scripts use the AWS Python SDK (Boto) under the hood. They are documented in the following section.

Prior to using these build commands, you must configure the project. Configuration parameters are split across JSON-formatted files located under the config/ directory. Configuration parameters are described in detail in an earlier section.

Build commands

This section describes important build commands and how to use them. If you want to use these commands right away to build the prototype, you may skip to the section titled "Deploy and run the prototype".

The packagelambda build command

Run this command to package the prototype's AWS Lambda functions and their dependencies (Image Processor and Frame Fetcher) into separate .zip packages (one per function). The deployment packages are created under the build/ directory.

pynt packagelambda # Package both functions and their dependencies into zip files.

pynt packagelambda[framefetcher] # Package only Frame Fetcher.

Currently, only Image Processor requires an external dependency, pytz. If you add features to Image Processor or Frame Fetcher that require external dependencies, you should install the dependencies using Pip by issuing the following command.

pip install <module-name> -t <path-to-project-dir>/lambda/<lambda-function-dir>

For example, let's say you want to perform image processing in the Image Processor Lambda function. You may decide on using the Pillow image processing library. To ensure Pillow is packaged with your Lambda function in one .zip file, issue the following command:

pip install Pillow -t <path-to-project-dir>/lambda/imageprocessor #Install Pillow dependency

You can find more details on installing AWS Lambda dependencies here.

The deploylambda build command

Run this command before you run createstack. The deploylambda command uploads Image Processor and Frame Fetcher .zip packages to Amazon S3 for pickup by AWS CloudFormation while creating the prototype's stack. This command will parse the deployment Amazon S3 bucket name and keys names from the cfn-params.json file. If the bucket does not exist, the script will create it. This bucket must be in the same AWS region as the AWS CloudFormation stack, or else the stack creation will fail. Without parameters, the command will deploy the .zip packages of both Image Processor and Frame Fetcher. You can specify either “imageprocessor” or “framefetcher” as a parameter between square brackets to deploy an individual function.

Here are sample command invocations.

pynt deploylambda # Deploy both functions to Amazon S3.

pynt deploylambda[framefetcher] # Deploy only Frame Fetcher to Amazon S3.

The createstack build command

The createstack command creates the prototype's AWS CloudFormation stack behind the scenes by invoking the create_stack() API. The AWS CloudFormation template used is located at aws-infra/aws-infra-cfn.yaml under the project’s root directory. The prototype's stack requires a number of parameters to be successfully created. The createstack script reads parameters from both global-params.json and cfn-params.json configuration files. The script then passes those parameters to the create_stack() call.

Note that you must, first, package and deploy Image Processor and Frame Fetcher functions to Amazon S3 using the packagelambda and deploylambda commands (documented later in this guid) for the AWS CloudFormation stack creation to succeed.

You can issue the command as follows:

pynt createstack

Stack creation should take only a couple of minutes. At any time, you can check on the prototype's stack status either through the AWS CloudFormation console or by issuing the following command.

pynt stackstatus

Congratulations! You’ve just created the prototype's entire architecture in your AWS account.

The deletestack build command

The deletestack command, once issued, does a few things. First, it empties the Amazon S3 bucket used to store video frame images. Next, it calls the AWS CloudFormation delete_stack() API to delete the prototype's stack from your account. Finally, it removes any unneeded resources not deleted by the stack (for example, the prototype's API Gateway Usage Plan resource).

You can issue the deletestack command as follows.

pynt deletestack

As with createstack, you can monitor the progress of stack deletion using the stackstatus build command.

The deletedata build command

The deletedata command, once issued, empties the Amazon S3 bucket used to store video frame images. Next, it also deletes all items in the DynamoDB table used to store frame metadata.

Use this command to clear all previously ingested video frames and associated metadata. The command will ask for confirmation [Y/N] before proceeding with deletion.

You can issue the deletedata command as follows.

pynt deletedata

The stackstatus build command

The stackstatus command will query AWS CloudFormation for the status of the prototype's stack. This command is most useful for quickly checking that the prototype is up and running (i.e. status is "CREATE_COMPLETE" or "UPDATE_COMPLETE") and ready to serve requests from the Web UI.

You can issue the command as follows.

pynt stackstatus # Get the prototype's Stack Status

The webui build command

Run this command when the prototype's stack has been created (using createstack). The webui command “builds” the Web UI through which you can monitor incoming captured video frames. First, the script copies the webui/ directory verbatim into the project’s build/ directory. Next, the script generates an apigw.js file which contains the API Gateway base URL and the API key to be used by Web UI for invoking the Fetch Frames function deployed in AWS Lambda. This file is created in the Web UI build directory.

You can issue the Web UI build command as follows.

pynt webui

The webuiserver build command

The webuiserver command starts a local, lightweight, Python-based HTTP server on your machine to serve Web UI from the build/web-ui/ directory. Use this command to serve the prototype's Web UI for development and demonstration purposes. You can specify the server’s port as pynt task parameter, between square brackets.

Here’s sample invocation of the command.

pynt webuiserver # Starts lightweight HTTP Server on port 8080.

The videocaptureip and videocapture build commands

The videocaptureip command fires up the MJPEG-based video capture client (source code under the client/ directory). This command accepts, as parameters, an MJPEG stream URL and an optional frame capture rate. The capture rate is defined as 1 every X number of frames. Captured frames are packaged, serialized, and sent to the Kinesis Frame Stream. The video capture client for IP cameras uses Open CV 3 to do simple image processing operations on captured frame images – mainly image rotation.

Here’s a sample command invocation.

pynt videocaptureip["http://192.168.0.2/video",20] # Captures 1 frame every 20.

On the other hand, the videocapture command (without the trailing 'ip'), fires up a video capture client that captures frames from a camera attached to the machine on which it runs. If you run this command on your laptop, for instance, the client will attempt to access its built-in video camera. This video capture client relies on Open CV 3 to capture video from physically connected cameras. Captured frames are packaged, serialized, and sent to the Kinesis Frame Stream.

Here’s a sample invocation.

pynt videocapture[20] # Captures one frame every 20.

Deploy and run the prototype

In this section, we are going use project's build commands to deploy and run the prototype in your AWS account. We’ll use the commands to create the prototype's AWS CloudFormation stack, build and serve the Web UI, and run the Video Cap client.

Prepare your development environment, and ensure configuration parameters are set as you wish.

On your machine, in a command line terminal change into the root directory of the project. Activate your virtual Python environment. Then, enter the following commands:

$ pynt packagelambda #First, package code & configuration files into .zip files

#Command output without errors

$ pynt deploylambda #Second, deploy your lambda code to Amazon S3

#Command output without errors

$ pynt createstack #Now, create the prototype's CloudFormation stack

#Command output without errors

$ pynt webui #Build the Web UI

#Command output without errors
  • On your machine, in a separate command line terminal:
$ pynt webuiserver #Start the Web UI server on port 8080 by default
  • In your browser, access http://localhost:8080 to access the prototype's Web UI. You should see a screen similar to this:

Empty Web UI

Now turn on your IP camera or launch the app on your smartphone. Ensure that your camera is accepting connections for streaming MJPEG video over HTTP, and identify the local URL for accessing that stream.

Then, in a terminal window at the root directory of the project, issue this command:

$ pynt videocaptureip["<your-ip-cam-mjpeg-url>",<capture-rate>]
  • Or, if you don’t have an IP camera and would like to use a built-in camera:
$ pynt videocapture[<frame-capture-rate>]
  • Few seconds after you execute this step, the dashed area in the Web UI will auto-populate with captured frames, side by side with labels recognized in them.

When you are done

After you are done experimenting with the prototype, perform the following steps to avoid unwanted costs.

  • Terminate video capture client(s) (press Ctrl+C in command line terminal where you got it running)
  • Close all open Web UI browser windows or tabs.
  • Execute the pynt deletestack command (see docs above)
  • After you run deletestack, visit the AWS CloudFormation console to double-check the stack is deleted.
  • Ensure that Amazon S3 buckets and objects within them are deleted.

Remember, you can always setup the entire prototype again with a few simple commands.

License

Licensed under the Amazon Software License.

A copy of the License is located at

http://aws.amazon.com/asl/

The AWS CloudFormation Stack (optional read)

Let’s quickly go through the stack that AWS CloudFormation sets up in your account based on the template. AWS CloudFormation uses as much parallelism as possible while creating resources. As a result, some resources may be created in an order different than what I’m going to describe here.

First, AWS CloudFormation creates the IAM roles necessary to allow AWS services to interact with one another. This includes the following.

ImageProcessorLambdaExecutionRole – a role to be assumed by the Image Processor lambda function. It allows full access to Amazon DynamoDB, Amazon S3, Amazon SNS, and AWS CloudWatch Logs. The role also allows read-only access to Amazon Kinesis and Amazon Rekognition. For simplicity, only managed AWS role permission policies are used.

FrameFetcherLambdaExecutionRole – a role to be assumed by the Frame Fetcher lambda function. It allows full access to Amazon S3, Amazon DynamoDB, and AWS CloudWatch Logs. For simplicity, only managed AWS permission policies are used. In parallel, AWS CloudFormation creates the Amazon S3 bucket to be used to store the captured video frame images. It also creates the Kinesis Frame Stream to receive captured video frame images from the Video Cap client.

Next, the Image Processor lambda function is created in addition to an AWS Lambda Event Source Mapping to allow Amazon Kinesis to trigger Image Processor once new captured video frames are available.

The Frame Fetcher lambda function is also created. Frame Fetcher is a simple lambda function that responds to a GET request by returning the latest list of frames, in descending order by processing timestamp, up to a configurable number of hours, called the “fetch horizon” (check the framefetcher-params.json file for more run-time configuration parameters). Necessary AWS Lambda Permissions are also created to permit Amazon API Gateway to invoke the Frame Fetcher lambda function.

AWS CloudFormation also creates the DynamoDB table where Enriched Frame metadata is stored by the Image Processor lambda function as described in the architecture overview section of this post. A Global Secondary Index (GSI) is also created; to be used by the Frame Fetcher lambda function in fetching Enriched Frame metadata in descending order by time of capture.

Finally, AWS CloudFormation creates the Amazon API Gateway resources necessary to allow the Web UI to securely invoke the Frame Fetcher lambda function with a GET request to a public API Gateway URL.

The following API Gateway resources are created.

REST API named “RtRekogRestAPI” by default.

An API Gateway resource with a path part set to “enrichedframe” by default.

A GET API Gateway method associated with the “enrichedframe” resource. This method is configured with Lambda proxy integration with the Frame Fetcher lambda function (learn more about AWS API Gateway proxy integration here). The method is also configured such that an API key is required.

An OPTIONS API Gateway method associated with the “enrichedframe” resource. This method’s purpose is to enable Cross-Origin Resource Sharing (CORS). Enabling CORS allows the Web UI to make Ajax requests to the Frame Fetcher API Gateway URL. Note that the Frame Fetcher lambda function must, itself, also return the Access-Control-Allow-Origin CORS header in its HTTP response.

A “development” API Gateway deployment to allow the invocation of the prototype's API over the Internet.

A “development” API Gateway stage for the API deployment along with an API Gateway usage plan named “development-plan” by default.

An API Gateway API key, name “DevApiKey” by default. The key is associated with the “development” stage and “development-plan” usage plan.

All defaults can be overridden in the cfn-params.json configuration file. That’s it for the prototype's AWS CloudFormation stack! This stack was designed primarily for development/demo purposes, especially how the Amazon API Gateway resources are set up.

FAQ

Q: Why is this project titled "amazon-rekognition-video-analyzer" despite the security-focused use case?

A: Although this prototype was conceived to address the security monitoring and alerting use case, you can use the prototype's architecture and code as a starting point to address a wide variety of use cases involving low-latency analysis of live video frames with Amazon Rekognition.

Download Details:
Author: aws-samples
Source Code: https://github.com/aws-samples/amazon-rekognition-video-analyzer
License: View license

#opencv  #python #aws 

What is GEEK

Buddha Community

Amazon Rekognition Video Analyzer Written in Opencv
The  NineHertz

The NineHertz

1630909606

How to Hire Video Game Developers For Video Game Development Company?

The gaming industry has taken a boom in the last few years. If we talk about numbers, according to NewZoo, the worth of the video gaming industry was $159.3 Billion in 2020. Video games are not just something for fun now, players and users expect much more from video game developers. Creating such products that just do not satisfy the player’s needs and exceed their expectations is what video game development company are thriving for.

Though kickstarting a new game-making studio is not an easy task. This business requires a team with a huge passion to create games and earn money from these video games. The idea of the approach is to create such unique games that will reach millions of people in the world and gain popularity. This growth demands more professionals in this field.

This just can not be obtained by finding someone with a good CV, the whole process includes a deep dig down to grab the right talent. Read on to learn more about Mobile game developers and the process of hiring video game developers.

Read Complete Blog Here -  https://theninehertz.com/blog/how-to-hire-video-game-developers-video-game-development


#Video Game Development

#Video Game developers

#Video game development studio

#Video game development services

 

Stuti .

Stuti .

1624862078

Top 10 Video Calling & Conferencing APIs & SDKs | Feb 2022

Top Video Calling API & SDK and Video Conferencing API & SDK for Developers are integrated and embeddable video calling API, and programmable live video Conferencing API, in just a few lines of code.

video calling api

A video calling feature can be easily integrated into any existing application, allowing to leverage the functionality in order to improve productivity. This is a great idea for saving money and development time. To do so all we need to do is integrate the API or SDK for video calling into the existing software.

In the rapidly modernizing world, technology has advanced in a hurtle. The use of mobile gadgets has surged a lot. The world is growing digital and everything is accessible from a single place. In the company of modernization, there is a huge increase in digitalization. Virtual communication has been nowadays an important tool to keep up with people. Due to virtual video communication, we can today talk face-to-face overseas too. Such a great invention to make work easy.

Many companies work on the aspect of virtual communication. The video calls we do to communicate with our clients, corporates, and family are developed with the help of Software Development Kits (SDKs). Companies align with competition and design their SDKs to support their video calling applications and websites. These days conferencing has become so efficient without any lags or errors or issues. This is just because of the effective product delivery of the companies who have landed themselves to make an effective approach towards real-time communication.

What is Video Calling API?

Video Calling API are app development kits that complement apps that support video calling features on their applications to connect their customers with the community. These SDKs are developed by companies with their effective creative SDK ideas, built with interfaces designed in a lucrative way to make them attract users.

The rise in remote working has also led to a rise in video conferencing in companies in both internal and external environments. To detail a video SDK, let us understand the significance to develop a clear idea of how SDK works. Video conferencing is considered to be one of the most important tools for business considering the pandemic situation, video SDK makes it efficient.

Video Calling API makes efficient use of resources. It helps to lower the costs in many direct and indirect ways. The tangible costs are cut. The designed SDKs help in eliminating costs with their integrated functions.

Video Calling  API helps in faster delivery, saving a lot of time. They help in creating video conferences for businesses with their integrated.

Video Calling  API are a stable platform build-up for apps with video conferencing. They make video conferencing flexible and accessible from any place and any time.

Video SDK allows conferencing on a large scale helping businesses achieve their desired objectives.

Video conferencing has become significant over time and for that reason, a strong SDK build-up is now an urge for each company. Being the product providers, the companies who build these SDKs look to deliver all their innovations in it so that the end customer finds it very much involved and attractive. An ideal Video SDK must have these features.

Video conferencing

The very basic feature a Video SDK must have is an effective video conferencing interface. It must be compatible for one-to-one communication as well as communication on a mass scale. This is the foremost feature to address while choosing a video-conferencing application.

Real-time chats and messaging

A video conferencing application must be designed to provide real-time chats in an ongoing meeting. This helps to supplement clarity during the meeting virtually, through multimedia channels.

Audio and video recording

In an ideal video conferencing, the users generally believe to have a backup of that communication for the future. This is the top-notch requirement for a company to build up a stable domain.

Sharing the screen

A screen share enables viewing access to the participants of the meeting, developing a clear perspective of ideas thought to the ideas delivered.
Push notifications
An attractive Video SDK must support enabling notifications while the conference to make discussions acknowledged at the right time. This helps in running a business conference smoothly.

Effective flexibility and scalability

A video SDK must provide effective scalability so that it can be addressed with any supportive device without a screen of less clarity. It must be flexible enough to support all the devices for its accessibility.
In the market of real-time communication, various companies offer customizable video SDK interfaces for their clients. Due to the current pandemic situation, the real-time communication industry has increased a lot. People have made virtual communication as the most used method to communicate, rather it is their professional or personal life. Many companies have launched their Video SDKs with customizable features.

Here are Top 10  Video Calling APIs & SDKs Provider Choose India, US, Canada

1. Video SDK Embed Live Video Calling & Conferencing API & SDK 

Video SDK is a web-RTC company that looks into creating lucrative Video SDKs and APIs for their clients. It looks for better engagement of their clients by supporting them with providing the best products for the end-users. Video SDK production, delivering sub 100ms to 150ms low latency streaming in the real-time community creating its image as the ideal platform for video conferencing due to its flexible and scalable SDKs. They aim at delivering some of the best experiences to their clients with their customizable SDKs.

2. ZujoNow Live Video Calling Solution for Business Collaboration

Zujonow is a company that develops its product on cutting-edge technologies. It delivers products for its clients based on video conferencing with effective scalability and customizable SDKs. They also deal in products like live streaming, on-demand videos, and real-time communication. Zujonow work is a well-crafted platform for education and other related industries.

3. EnableX Live Video Calling API for Web & App

EnableX works on the development of communication APIs. it focuses on communication solutions and provides services in real-time communication for its clients. It delivers its products with SMS and chat interfaces too. EnableX works on developing educational APIs for students and also maintaining portfolios.

4. Daily.co Powerful Live Video Calling API

Daily.co is a real-time audio and video API developer, working on focusing on the best scalable video conferencing api for its clients. Daily.co works on developing global infrastructure delivering the best call quality on a timely basis, considering their web-RTC to be a source of service to the clients.

5. Eyeson Live Video Calling Solution for Business Collaboration

Eyeson masters into high-performing API including managed hosting and scaling for web-based business workflows on any device. It has a patented single Stream Technology merges any live media, data and participants in real-time into a single video and audio stream. The cloud services at eyeson can immediately be used for world wide scaling. It provides a world-class facility for its clients with guaranteed privacy.

6. Agora  Live video calling api

Agora.io is a web real-time communication company that develops SDKs and APIs. It works on engagement for the users by real-time voice, video call javascript  messaging, and live streaming products. They also have set up classrooms for students for learning with an interactive whiteboard.

7. Twilio cloud-based Live Video Calling APIs and SDK

Twilio develops video applications that are fully customizable, scalable, and flexible for usage. It constructs applications and connectivity and has its build-up. It makes channels for video, chats, and programmable chats. Twilio also looks at SMS build-up for its clients. It provides solutions based on real-time communication and scalability and video calling api 

8. PubNub Live Video Calling and Realtime Communication Platform

Pubnub is an in-app chat for real-time chat engagement. It retains full control, functionality, and customization without the time and expense of building in-house. It provides outsourcing to clients with the products like custom chat, effortless scalability, in-class integrations, and Chat UI support. They have a strong research window that looks for developing APIs for their clients.

9. Cometchat  Live Video Calling api Provider

Cometchat is designed for providing APIs and SDKs for various solutions for ed-tech, healthcare, dating, and social community. It is also devised for on-demand videos and live streaming. It allows its users to customize their Whitelabel as per their needs to make it feel like ownership. Cometchat is adaptive to all languages and has effective work data too.

10. Sinch  WebRTC- Live Video Conferencing API Solution

Sinch works on managing different APIs for messaging and calling. It puts forward the products for video calling, voice calls, SMS verification, and other engagement platforms. It provides solutions to different industries like health, retail, telecommunications, media and entertainment, and more. It provided operators opportunities for monetizing wholesale, preventing fraud, and other activities.

11. Apphitect  video and chat calling api

Apphitect focuses on mobile app development for android and iOS. It also engages its clients with different solutions concerning messaging. Apphitect delivers app testings and mobile. It develops everything from wireframe to pixel. Apphitact is currently working in 40+ cities with its headquarters in UAE.

12. Vidyo audio and Live Video Calling API providers

Vidyo provides solutions to services like Branding and white-labeling and hybrid cloud expansion. It also works with solutions for deployment services and project management services. It works for several industries like health, education, government, finance, retail; and more. It promotes connectivity and engagement for the users and also focuses on video conferencing systems for businesses.

 

 

Due to the current pandemic, real-time communication has took a massive hype for its flexible availability. All the businesses, corporates, schools, and organisations had to run over the web. Even earlier video conferencing was an all time favorable option for corporates to abide with communicating with clients, over a distance. But in the latter period, recently, the whole world has become dependent on it. For the same cause, it led to emergence of various innovative ideas for bringing people closer together even at a distance. Video conferencing has today become vital and above all it is appreciable for the companies who have invested in bringing up ideas by developing customisable video SDKs for their clients to promote belongingness. The APIs and SDKs designed by the companies have made it easier to use by the end consumers too. Overall, video conferencing makes work flexible and accessible to all, making itself categorise into an principal element of businesses.

#video-conferencing-api#video-sdk #videoapi #video-conferencing #video-sdk-comparison  #video-calling-api

Marcus Anthony

1612355729

Amazon Pay Clone, Amazon Pay Clone Script, Recharge & Wallet App Solution

Mobile wallet applications have become the new trend in today’s world. Apps like Amazon Pay, Paytm, PayPal are some of the leading apps that are owned and used by millions. Be it paying bills, recharging, or money transactions, everything has turned easier because of these apps. There were days when people used to travel for hours to do these tasks have been totally transformed. Moreover, consumers can use these e-wallet apps while paying in a store, either for shopping or while eating out. Thus, as far as mobile wallets are concerned, they are a convenient way for handling all the tasks involving finance.

As an aspiring entrepreneur, if you wish to succeed in your business, without second thoughts, go for Amazon Pay clone app development. Let’s narrow down your thinking processes for a quicker stride forward by analyzing the types of apps first.

Types of e-wallet apps you could develop:

Retail application: An app like Amazon is considered the retail app because it has a mobile wallet in it. It has all the basic functionalities, which helps users to redeem coupons and reward points. All the payment modes are accessible through the app, including net banking.

Dedicated app: The app allows P2P money transactions by storing a variety of cards. You could also make international money transfers using this app. Example: PayPal, Apple Pay, and Amazon Pay.

PoS payments: The PoS payment wallet apps are found at the stores. It is exclusively used by the users to make contactless payments without having to stand in a long queue.

Wrapping up,
Choose the best type of e-wallet app you want to develop and join forces with our Appdupe. Grab the cutting-edge Amazon Pay Clone script and launch an app in a week!

##amazon pay clone ##amazon pay clone script ##amazon pay clone app ##amazon pay clone app development ##amazon pay app clone ##amazon pay app clone development

studio52 dubai

studio52 dubai

1621769539

How to find the best video production company in Dubai?

How to find the best video production company in Dubai?We are the best video production company in Dubai, UAE. We offer Corporate Video, event video, animation video, safety video and timelapse video in most engaging and creative ways.

#video production company #video production dubai #video production services #video production services dubai #video production #video production house

Amazon Rekognition Video Analyzer Written in Opencv

Create a Serverless Pipeline for Video Frame Analysis and Alerting

Introduction

Imagine being able to capture live video streams, identify objects using deep learning, and then trigger actions or notifications based on the identified objects -- all with low latency and without a single server to manage.

This is exactly what this project is going to help you accomplish with AWS. You will be able to setup and run a live video capture, analysis, and alerting solution prototype.

The prototype was conceived to address a specific use case, which is alerting based on a live video feed from an IP security camera. At a high level, the solution works as follows. A camera surveils a particular area, streaming video over the network to a video capture client. The client samples video frames and sends them over to AWS, where they are analyzed and stored along with metadata. If certain objects are detected in the analyzed video frames, SMS alerts are sent out. Once a person receives an SMS alert, they will likely want to know what caused it. For that, sampled video frames can be monitored with low latency using a web-based user interface.

Here's the prototype's conceptual architecture:

Architecture

Let's go through the steps necessary to get this prototype up and running. If you are starting from scratch and are not familiar with Python, completing all steps can take a few hours.

Preparing your development environment

Here’s a high-level checklist of what you need to do to setup your development environment.

  1. Sign up for an AWS account if you haven't already and create an Administrator User. The steps are published here.
  2. Ensure that you have Python 2.7+ and Pip on your machine. Instructions for that varies based on your operating system and OS version.
  3. Create a Python virtual environment for the project with Virtualenv. This helps keep project’s python dependencies neatly isolated from your Operating System’s default python installation. Once you’ve created a virtual python environment, activate it before moving on with the following steps.
  4. Use Pip to install AWS CLI. Configure the AWS CLI. It is recommended that the access keys you configure are associated with an IAM User who has full access to the following:
  • Amazon S3
  • Amazon DynamoDB
  • Amazon Kinesis
  • AWS Lambda
  • Amazon CloudWatch and CloudWatch Logs
  • AWS CloudFormation
  • Amazon Rekognition
  • Amazon SNS
  • Amazon API Gateway
  • Creating IAM Roles

The IAM User can be the Administrator User you created in Step 1.

5.   Make sure you choose a region where all of the above services are available. Regions us-east-1 (N. Virginia), us-west-2 (Oregon), and eu-west-1 (Ireland) fulfill this criterion. Visit this page to learn more about service availability in AWS regions.

6.   Use Pip to install Open CV 3 python dependencies and then compile, build, and install Open CV 3 (required by Video Cap clients). You can follow this guide to get Open CV 3 up and running on OS X Sierra with Python 2.7. There's another guide for Open CV 3 and Python 3.5 on OS X Sierra. Other guides exist as well for Windows and Raspberry Pi.

7.   Use Pip to install Boto3. Boto is the Amazon Web Services (AWS) SDK for Python, which allows Python developers to write software that makes use of Amazon services like S3 and EC2. Boto provides an easy to use, object-oriented API as well as low-level direct access to AWS services.

8.   Use Pip to install Pynt. Pynt enables you to write project build scripts in Python.

9.   Clone this GitHub repository. Choose a directory path for your project that does not contain spaces (I'll refer to the full path to this directory as <path-to-project-dir>).

10.   Use Pip to install pytz. Pytz is needed for timezone calculations. Use the following commands:

pip install pytz # Install pytz in your virtual python env

pip install pytz -t <path-to-project-dir>/lambda/imageprocessor/ # Install pytz to be packaged and deployed with the Image Processor lambda function

Finally, obtain an IP camera. If you don’t have an IP camera, you can use your smartphone with an IP camera app. This is useful in case you want to test things out before investing in an IP camera. Also, you can simply use your laptop’s built-in camera or a connected USB camera. If you use an IP camera, make sure your camera is connected to the same Local Area Network as the Video Capture client.

Configuring the project

In this section, I list every configuration file, parameters within it, and parameter default values. The build commands detailed later extract the majority of their parameters from these configuration files. Also, the prototype's two AWS Lambda functions - Image Processor and Frame Fetcher - extract parameters at runtime from imageprocessor-params.json and framefetcher-params.json respectively.

NOTE: Do not remove any of the attributes already specified in these files.

NOTE: You must set the value of any parameter that has the tag NO-DEFAULT

config/global-params.json

Specifies “global” build configuration parameters. It is read by multiple build scripts.

{
    "StackName" : "video-analyzer-stack"
}

Parameters:

  • StackName - The name of the stack to be created in your AWS account.

config/cfn-params.json

Specifies and overrides default values of AWS CloudFormation parameters defined in the template (located at aws-infra/aws-infra-cfn.yaml). This file is read by a number of build scripts, including createstack, deploylambda, and webui.

{
    "SourceS3BucketParameter" : "<NO-DEFAULT>",
    "ImageProcessorSourceS3KeyParameter" : "src/lambda_imageprocessor.zip",
    "FrameFetcherSourceS3KeyParameter" : "src/lambda_framefetcher.zip",

    "FrameS3BucketNameParameter" : "<NO-DEFAULT>",

    "FrameFetcherApiResourcePathPart" : "enrichedframe",
    "ApiGatewayRestApiNameParameter" : "VidAnalyzerRestApi",
    "ApiGatewayStageNameParameter": "development",
    "ApiGatewayUsagePlanNameParameter" : "development-plan"
}

Parameters:

SourceS3BucketParameter - The Amazon S3 bucket to which your AWS Lambda function packages (.zip files) will be deployed. If a bucket with such a name does not exist, the deploylambda build command will create it for you with appropriate permissions. AWS CloudFormation will access this bucket to retrieve the .zip files for Image Processor and Frame Fetcher AWS Lambda functions.

ImageProcessorSourceS3KeyParameter - The Amazon S3 key under which the Image Processor function .zip file will be stored.

FrameFetcherSourceS3KeyParameter - The Amazon S3 key under which the Frame Fetcher function .zip file will be stored.

FrameS3BucketNameParameter - The Amazon S3 bucket that will be used for storing video frame images. There must not be an existing S3 bucket with the same name.

FrameFetcherApiResourcePathPart - The name of the Frame Fetcher API resource path part in the API Gateway URL.

ApiGatewayRestApiNameParameter - The name of the API Gateway REST API to be created by AWS CloudFormation.

ApiGatewayStageNameParameter - The name of the API Gateway stage to be created by AWS CloudFormation.

ApiGatewayUsagePlanNameParameter - The name of the API Gateway usage plan to be created by AWS CloudFormation.

config/imageprocessor-params.json

Specifies configuration parameters to be used at run-time by the Image Processor lambda function. This file is packaged along with the Image Processor lambda function code in a single .zip file using the packagelambda build script.

{
    "s3_bucket" : "<NO-DEFAULT>",
    "s3_key_frames_root" : "frames/",

    "ddb_table" : "EnrichedFrame",

    "rekog_max_labels" : 123,
    "rekog_min_conf" : 50.0,

    "label_watch_list" : ["Human", "Pet", "Bag", "Toy"],
    "label_watch_min_conf" : 90.0,
    "label_watch_phone_num" : "",
    "label_watch_sns_topic_arn" : "",
    "timezone" : "US/Eastern"
}

s3_bucket - The Amazon S3 bucket in which Image Processor will store captured video frame images. The value specified here must match the value specified for the FrameS3BucketNameParameter parameter in the cfn-params.json file.

s3_key_frames_root - The Amazon S3 key prefix that will be prepended to the keys of all stored video frame images.

ddb_table - The Amazon DynamoDB table in which Image Processor will store video frame metadata. The default value,EnrichedFrame, matches the default value of the AWS CloudFormation template parameter DDBTableNameParameter in the aws-infra/aws-infra-cfn.yaml template file.

rekog_max_labels - The maximum number of labels that Amazon Rekognition can return to Image Processor.

rekog_min_conf - The minimum confidence required for a label identified by Amazon Rekognition. Any labels with confidence below this value will not be returned to Image Processor.

label_watch_list - A list of labels for to watch out for. If any of the labels specified in this parameter are returned by Amazon Rekognition, an SMS alert will be sent via Amazon SNS. The label's confidence must exceed label_watch_min_conf.

label_watch_min_conf - The minimum confidence required for a label to trigger a Watch List alert.

label_watch_phone_num - The mobile phone number to which a Watch List SMS alert will be sent. Does not have a default value. You must configure a valid phone number adhering to the E.164 format (e.g. +1404XXXYYYY) for the Watch List feature to become active.

label_watch_sns_topic_arn - The SNS topic ARN to which you want Watch List alert messages to be sent. The alert message contains a notification text in addition to a JSON formatted list of Watch List labels found. This can be used to publish alerts to any SNS subscribers, such as Amazon SQS queues.

timezone - The timezone used to report time and date in SMS alerts. By default, it is "US/Eastern". See this list of country codes, names, continents, capitals, and pytz timezones).

config/framefetcher-params.json

Specifies configuration parameters to be used at run-time by the Frame Fetcher lambda function. This file is packaged along with the Frame Fetcher lambda function code in a single .zip file using the packagelambda build script.

{
    "s3_pre_signed_url_expiry" : 1800,

    "ddb_table" : "EnrichedFrame",
    "ddb_gsi_name" : "processed_year_month-processed_timestamp-index",

    "fetch_horizon_hrs" : 24,
    "fetch_limit" : 3
}

s3_pre_signed_url_expiry - Frame Fetcher returns video frame metadata. Along with the returned metadata, Frame Fetcher generates and returns a pre-signed URL for every video frame. Using a pre-signed URL, a client (such as the Web UI) can securely access the JPEG image associated with a particular frame. By default, the pre-signed URLs expire in 30 minutes.

ddb_table - The Amazon DynamoDB table from which Frame Fetcher will fetch video frame metadata. The default value,EnrichedFrame, matches the default value of the AWS CloudFormation template parameter DDBTableNameParameter in the aws-infra/aws-infra-cfn.yaml template file.

ddb_gsi_name - The name of the Amazon DynamoDB Global Secondary Index that Frame Fetcher will use to query frame metadata. The default value matches the default value of the AWS CloudFormation template parameter DDBGlobalSecondaryIndexNameParameter in the aws-infra/aws-infra-cfn.yaml template file.

fetch_horizon_hrs - Frame Fetcher will exclude any video frames that were ingested prior to the point in the past represented by (time now - fetch_horizon_hrs).

fetch_limit - The maximum number of video frame metadata items that Frame Fetcher will retrieve from Amazon DynamoDB.

Building the prototype

Common interactions with the project have been simplified for you. Using pynt, the following tasks are automated with simple commands:

  • Creating, deleting, and updating the AWS infrastructure stack with AWS CloudFormation
  • Packaging lambda code into .zip files and deploying them into an Amazon S3 bucket
  • Running the video capture client to stream from a built-in laptop webcam or a USB camera
  • Running the video capture client to stream from an IP camera (MJPEG stream)
  • Build a simple web user interface (Web UI)
  • Run a lightweight local HTTP server to serve Web UI for development and demo purposes

For a list of all available tasks, enter the following command in the root directory of this project:

pynt -l

The output represents the list of build commands available to you:

pynt -l output

Build commands are implemented as python scripts in the file build.py. The scripts use the AWS Python SDK (Boto) under the hood. They are documented in the following section.

Prior to using these build commands, you must configure the project. Configuration parameters are split across JSON-formatted files located under the config/ directory. Configuration parameters are described in detail in an earlier section.

Build commands

This section describes important build commands and how to use them. If you want to use these commands right away to build the prototype, you may skip to the section titled "Deploy and run the prototype".

The packagelambda build command

Run this command to package the prototype's AWS Lambda functions and their dependencies (Image Processor and Frame Fetcher) into separate .zip packages (one per function). The deployment packages are created under the build/ directory.

pynt packagelambda # Package both functions and their dependencies into zip files.

pynt packagelambda[framefetcher] # Package only Frame Fetcher.

Currently, only Image Processor requires an external dependency, pytz. If you add features to Image Processor or Frame Fetcher that require external dependencies, you should install the dependencies using Pip by issuing the following command.

pip install <module-name> -t <path-to-project-dir>/lambda/<lambda-function-dir>

For example, let's say you want to perform image processing in the Image Processor Lambda function. You may decide on using the Pillow image processing library. To ensure Pillow is packaged with your Lambda function in one .zip file, issue the following command:

pip install Pillow -t <path-to-project-dir>/lambda/imageprocessor #Install Pillow dependency

You can find more details on installing AWS Lambda dependencies here.

The deploylambda build command

Run this command before you run createstack. The deploylambda command uploads Image Processor and Frame Fetcher .zip packages to Amazon S3 for pickup by AWS CloudFormation while creating the prototype's stack. This command will parse the deployment Amazon S3 bucket name and keys names from the cfn-params.json file. If the bucket does not exist, the script will create it. This bucket must be in the same AWS region as the AWS CloudFormation stack, or else the stack creation will fail. Without parameters, the command will deploy the .zip packages of both Image Processor and Frame Fetcher. You can specify either “imageprocessor” or “framefetcher” as a parameter between square brackets to deploy an individual function.

Here are sample command invocations.

pynt deploylambda # Deploy both functions to Amazon S3.

pynt deploylambda[framefetcher] # Deploy only Frame Fetcher to Amazon S3.

The createstack build command

The createstack command creates the prototype's AWS CloudFormation stack behind the scenes by invoking the create_stack() API. The AWS CloudFormation template used is located at aws-infra/aws-infra-cfn.yaml under the project’s root directory. The prototype's stack requires a number of parameters to be successfully created. The createstack script reads parameters from both global-params.json and cfn-params.json configuration files. The script then passes those parameters to the create_stack() call.

Note that you must, first, package and deploy Image Processor and Frame Fetcher functions to Amazon S3 using the packagelambda and deploylambda commands (documented later in this guid) for the AWS CloudFormation stack creation to succeed.

You can issue the command as follows:

pynt createstack

Stack creation should take only a couple of minutes. At any time, you can check on the prototype's stack status either through the AWS CloudFormation console or by issuing the following command.

pynt stackstatus

Congratulations! You’ve just created the prototype's entire architecture in your AWS account.

The deletestack build command

The deletestack command, once issued, does a few things. First, it empties the Amazon S3 bucket used to store video frame images. Next, it calls the AWS CloudFormation delete_stack() API to delete the prototype's stack from your account. Finally, it removes any unneeded resources not deleted by the stack (for example, the prototype's API Gateway Usage Plan resource).

You can issue the deletestack command as follows.

pynt deletestack

As with createstack, you can monitor the progress of stack deletion using the stackstatus build command.

The deletedata build command

The deletedata command, once issued, empties the Amazon S3 bucket used to store video frame images. Next, it also deletes all items in the DynamoDB table used to store frame metadata.

Use this command to clear all previously ingested video frames and associated metadata. The command will ask for confirmation [Y/N] before proceeding with deletion.

You can issue the deletedata command as follows.

pynt deletedata

The stackstatus build command

The stackstatus command will query AWS CloudFormation for the status of the prototype's stack. This command is most useful for quickly checking that the prototype is up and running (i.e. status is "CREATE_COMPLETE" or "UPDATE_COMPLETE") and ready to serve requests from the Web UI.

You can issue the command as follows.

pynt stackstatus # Get the prototype's Stack Status

The webui build command

Run this command when the prototype's stack has been created (using createstack). The webui command “builds” the Web UI through which you can monitor incoming captured video frames. First, the script copies the webui/ directory verbatim into the project’s build/ directory. Next, the script generates an apigw.js file which contains the API Gateway base URL and the API key to be used by Web UI for invoking the Fetch Frames function deployed in AWS Lambda. This file is created in the Web UI build directory.

You can issue the Web UI build command as follows.

pynt webui

The webuiserver build command

The webuiserver command starts a local, lightweight, Python-based HTTP server on your machine to serve Web UI from the build/web-ui/ directory. Use this command to serve the prototype's Web UI for development and demonstration purposes. You can specify the server’s port as pynt task parameter, between square brackets.

Here’s sample invocation of the command.

pynt webuiserver # Starts lightweight HTTP Server on port 8080.

The videocaptureip and videocapture build commands

The videocaptureip command fires up the MJPEG-based video capture client (source code under the client/ directory). This command accepts, as parameters, an MJPEG stream URL and an optional frame capture rate. The capture rate is defined as 1 every X number of frames. Captured frames are packaged, serialized, and sent to the Kinesis Frame Stream. The video capture client for IP cameras uses Open CV 3 to do simple image processing operations on captured frame images – mainly image rotation.

Here’s a sample command invocation.

pynt videocaptureip["http://192.168.0.2/video",20] # Captures 1 frame every 20.

On the other hand, the videocapture command (without the trailing 'ip'), fires up a video capture client that captures frames from a camera attached to the machine on which it runs. If you run this command on your laptop, for instance, the client will attempt to access its built-in video camera. This video capture client relies on Open CV 3 to capture video from physically connected cameras. Captured frames are packaged, serialized, and sent to the Kinesis Frame Stream.

Here’s a sample invocation.

pynt videocapture[20] # Captures one frame every 20.

Deploy and run the prototype

In this section, we are going use project's build commands to deploy and run the prototype in your AWS account. We’ll use the commands to create the prototype's AWS CloudFormation stack, build and serve the Web UI, and run the Video Cap client.

Prepare your development environment, and ensure configuration parameters are set as you wish.

On your machine, in a command line terminal change into the root directory of the project. Activate your virtual Python environment. Then, enter the following commands:

$ pynt packagelambda #First, package code & configuration files into .zip files

#Command output without errors

$ pynt deploylambda #Second, deploy your lambda code to Amazon S3

#Command output without errors

$ pynt createstack #Now, create the prototype's CloudFormation stack

#Command output without errors

$ pynt webui #Build the Web UI

#Command output without errors
  • On your machine, in a separate command line terminal:
$ pynt webuiserver #Start the Web UI server on port 8080 by default
  • In your browser, access http://localhost:8080 to access the prototype's Web UI. You should see a screen similar to this:

Empty Web UI

Now turn on your IP camera or launch the app on your smartphone. Ensure that your camera is accepting connections for streaming MJPEG video over HTTP, and identify the local URL for accessing that stream.

Then, in a terminal window at the root directory of the project, issue this command:

$ pynt videocaptureip["<your-ip-cam-mjpeg-url>",<capture-rate>]
  • Or, if you don’t have an IP camera and would like to use a built-in camera:
$ pynt videocapture[<frame-capture-rate>]
  • Few seconds after you execute this step, the dashed area in the Web UI will auto-populate with captured frames, side by side with labels recognized in them.

When you are done

After you are done experimenting with the prototype, perform the following steps to avoid unwanted costs.

  • Terminate video capture client(s) (press Ctrl+C in command line terminal where you got it running)
  • Close all open Web UI browser windows or tabs.
  • Execute the pynt deletestack command (see docs above)
  • After you run deletestack, visit the AWS CloudFormation console to double-check the stack is deleted.
  • Ensure that Amazon S3 buckets and objects within them are deleted.

Remember, you can always setup the entire prototype again with a few simple commands.

License

Licensed under the Amazon Software License.

A copy of the License is located at

http://aws.amazon.com/asl/

The AWS CloudFormation Stack (optional read)

Let’s quickly go through the stack that AWS CloudFormation sets up in your account based on the template. AWS CloudFormation uses as much parallelism as possible while creating resources. As a result, some resources may be created in an order different than what I’m going to describe here.

First, AWS CloudFormation creates the IAM roles necessary to allow AWS services to interact with one another. This includes the following.

ImageProcessorLambdaExecutionRole – a role to be assumed by the Image Processor lambda function. It allows full access to Amazon DynamoDB, Amazon S3, Amazon SNS, and AWS CloudWatch Logs. The role also allows read-only access to Amazon Kinesis and Amazon Rekognition. For simplicity, only managed AWS role permission policies are used.

FrameFetcherLambdaExecutionRole – a role to be assumed by the Frame Fetcher lambda function. It allows full access to Amazon S3, Amazon DynamoDB, and AWS CloudWatch Logs. For simplicity, only managed AWS permission policies are used. In parallel, AWS CloudFormation creates the Amazon S3 bucket to be used to store the captured video frame images. It also creates the Kinesis Frame Stream to receive captured video frame images from the Video Cap client.

Next, the Image Processor lambda function is created in addition to an AWS Lambda Event Source Mapping to allow Amazon Kinesis to trigger Image Processor once new captured video frames are available.

The Frame Fetcher lambda function is also created. Frame Fetcher is a simple lambda function that responds to a GET request by returning the latest list of frames, in descending order by processing timestamp, up to a configurable number of hours, called the “fetch horizon” (check the framefetcher-params.json file for more run-time configuration parameters). Necessary AWS Lambda Permissions are also created to permit Amazon API Gateway to invoke the Frame Fetcher lambda function.

AWS CloudFormation also creates the DynamoDB table where Enriched Frame metadata is stored by the Image Processor lambda function as described in the architecture overview section of this post. A Global Secondary Index (GSI) is also created; to be used by the Frame Fetcher lambda function in fetching Enriched Frame metadata in descending order by time of capture.

Finally, AWS CloudFormation creates the Amazon API Gateway resources necessary to allow the Web UI to securely invoke the Frame Fetcher lambda function with a GET request to a public API Gateway URL.

The following API Gateway resources are created.

REST API named “RtRekogRestAPI” by default.

An API Gateway resource with a path part set to “enrichedframe” by default.

A GET API Gateway method associated with the “enrichedframe” resource. This method is configured with Lambda proxy integration with the Frame Fetcher lambda function (learn more about AWS API Gateway proxy integration here). The method is also configured such that an API key is required.

An OPTIONS API Gateway method associated with the “enrichedframe” resource. This method’s purpose is to enable Cross-Origin Resource Sharing (CORS). Enabling CORS allows the Web UI to make Ajax requests to the Frame Fetcher API Gateway URL. Note that the Frame Fetcher lambda function must, itself, also return the Access-Control-Allow-Origin CORS header in its HTTP response.

A “development” API Gateway deployment to allow the invocation of the prototype's API over the Internet.

A “development” API Gateway stage for the API deployment along with an API Gateway usage plan named “development-plan” by default.

An API Gateway API key, name “DevApiKey” by default. The key is associated with the “development” stage and “development-plan” usage plan.

All defaults can be overridden in the cfn-params.json configuration file. That’s it for the prototype's AWS CloudFormation stack! This stack was designed primarily for development/demo purposes, especially how the Amazon API Gateway resources are set up.

FAQ

Q: Why is this project titled "amazon-rekognition-video-analyzer" despite the security-focused use case?

A: Although this prototype was conceived to address the security monitoring and alerting use case, you can use the prototype's architecture and code as a starting point to address a wide variety of use cases involving low-latency analysis of live video frames with Amazon Rekognition.

Download Details:
Author: aws-samples
Source Code: https://github.com/aws-samples/amazon-rekognition-video-analyzer
License: View license

#opencv  #python #aws