• About the Project
  • Motivation
  • Built With
  • Roadmap
• Implementation
  • Local implementation
  • Gradle implementation
• Usage
  • Basic Usage
  • Parameters
  • Examples
    • Image demo
    • Real time demo

Pidroid is android library to perform fast face detection, pupil detection and landmarks detection without third-party dependencies dependencies.

This library has been developed to facilitate the integration of a face detection method in your Android application.

Usually, to install a face detection library, some integration with third-party software such as OpenCV or Tensorflow Lite or registration on a platform such as Firebase is necessary.

On the other hand, the weight of face detection models is usually high, which causes the application to grow unnecessarily.

Pidroid provides a completely dependency-free alternative to make it easier for developers to use these libraries.

This library currently is full developed in Kotlin with C++, no additional dependencies or installation is needed :)

• Face detection feature is based on Pixel Intensity Comparison-based Object detection paper (https://arxiv.org/pdf/1305.4537.pdf)
• Pupil detection feature is based on Eye pupil localization with an ensemble of randomized trees
• Landmarks detection feature is based on Fast Localization of Facial Landmark Points

• High speed processing
• Does not require additional dependencies
• There is no need for image preprocessing prior detection
• Face detection
• Pupil detection
• Landmarks detection
• Pure Java Pidroid (to delete avoid Kotlin dependency)
• Publish Library as Github Package
• Rotated faces detection
• Neon support
• SSE support

First, right click on your app module and click on Open Module Settings.

Now, click on + button above app module. The following window will open:

Then click on Import .JAR/.AAR Package and select pidroid-release.aar module. Your project look like this:

Finally, add the following line to your build.gradle of the app module:

implementation project(path: ':pidroid-release')

Your app's build.gradle look like this:

Still in progress..

Pidroid library provides singleton object called "Pidroid". First, you need setup library in your Activity onCreate.

override fun onCreate(savedInstanceState: Bundle?) {
        super.onCreate(savedInstanceState)
        setContentView(...)

		// Do some stuff...

        val pidroidConfig = PidroidConfig()
        Pidroid.setup(this, pidroidConfig)
    }

After setup finish, you can call detectFace function like this (example with Bitmap):

fun callDetectFace(bitmap: Bitmap): FaceDetectionResult {
        val pixels = IntArray(bitmap.width * bitmap.height * 4)
        bitmap.getPixels(pixels, 0, bitmap.width, 0, 0, bitmap.width, bitmap.height)
        var dInfo = FaceDetectionResult()
        Pidroid.detectFace(pixels, bitmap.width, bitmap.height, dInfo)
        return dInfo
    }

All configuration parameters are in PidroidConfig class.

In this project you will find two sample applications. The first is a simple application that performs the prediction on a bitmap and draws the result.

The second opens the camera and performs face, pupil and landmark detection in real time using a separate Thread so as not to block the UI.

Still in progress..

Still in progress..

• Adrián Suárez Parra (https://linkedin.com/in/adriansp3)

Copyright © 2021 Adrián Suárez Parra (Suaro)

This software is distributed under the MIT license. See the LICENSE file for the full license text.