BlazePose for Full Body Keypoints Extraction.

Blazepose for Full Body Keypoints Extraction

It is worth mentioning that currently, Jan 2021, the full body is only available on Android and iOS through ML Kit API.

  • First, download and set up Integrated Development Environment (IDE) for Android app development .
  • Second, install some components such as SDKS and Android Emulator on your IDE that you will need later on.
  • Third , download “ML kit API” code of their quick start app in Android.
  • Fourth ,go to the file where you can run the app.
  • Fifth, create your Android Virtual Device to run the app on . This step involves many stages where you need to : Choose your device definition , determine your virtual device’s hardware capabilities such as camera, select a system image , and finally verify the configuration .
  • Sixth , edit your “run/debug configuration” .
  • Seventh , pick your Android device and hit “run” button.
  • Eighth, dealing with ml-kit interface.
  • Finally, showing BlazePose’s results .

Now let’s elaborate on every step

For the first step :

Android Studio Components’ installation

For the second step:

  • Android SDK Build-Tools
  • Android Emulator
  • Android SDK Platform-Tools
  • Android SDK Tools
  • Intel X86 Emulator Accelerator

The following figures show where these orders lie in Android Studio Interface

GO to tools ,then click on SDK Manager
Click on Android SDK , then pick SDK Tools, select the highlighted components , finally click OK.
Now switch to SDK Platforms , you may want to pick “Android 7.1.1(Nougat)” to be your SDK Platforms

For the third step :

For the fourth step :

For the fifth step :

Now you need to create the Android virtual device (AVD) you will run your app on. Remember that ml kit is a mobile SDK. As a result, if you want to run it on your pc/laptop, you will need to install Android Emulator as we did in the second step as well as creating virtual android device with capabilities that allow you to test the app on.

  1. Create your virtual device
  2. Choose your device definition
  3. Determine your virtual device’s hardware capabilities
  4. Select a system image
  5. Finally verify the configuration .
Creating virtual device
Create Virtual Device
Device Definition
  • Back-Facing Camera — where the lens faces away from the user.
  • Front-Facing Camera —where the lens faces toward the user.
Configure Hardware Profile

Tip :If you are willing to understand more about Android Virtual Device aka understanding why we are doing these steps , I highly recommend this resource.

x86 Images
  • Recommended
  • x86 Images
  • Other Images
Verifying AVD Configurations

“Congratulations for Creating your Android Virtual Device”

For the Sixth step :

Appropriate Run/Debug Configurations

For the seventh step:

Pick the Name of AVD
Run the app

For the Eighth Step:

let’s deal with ML Kit‘s easy-to-use interface

Left Figure: ML Kit interface
Right Figure : Infer the model on static images, choose the second option
Left Figure : Pose Detection
Right Figure : Choose Select image from album

Congratulations…..You have Earned it .. Here is the Final step where we view the Results

Apply the model on the image.

BlazePose’s Results :

BlazePose’s Results


“Finally, Clapping is your way to let us know you have liked our Article”




Machine Learning Engineer at VRapeutic . Alumna from University of Science and Technology at Zewail City.

Love podcasts or audiobooks? Learn on the go with our new app.

Recommended from Medium

Set A Image As A Wallpaper In Flutter

Feature Components in Android

Where can I find my Discord ID?

DIY Gradle build optimization

detekt, A static code analysis linter tool for Kotlin Android projects!

How to validate and debug Android app Analytics implementation using Firebase Debug View?

Android JetPack: Navigation Architecture Component

Android, Aligning Views In a Circle Shape

Get the Medium app

Alaa Hesham

Alaa Hesham

Machine Learning Engineer at VRapeutic . Alumna from University of Science and Technology at Zewail City.

More from Medium

Cab Fare Price Prediction

Does Dalle 2 understands abstraction?


Fast detection of fake and tampered cards with artificial intelligence