Huawei ML Kit -Text-Image Super Resolution

Neha Jeswani
Huawei Developers
Published in
4 min readFeb 19, 2021

--

Huawei ML Kit offers to overcome such challenges of our day to day life.

Huawei ML Kit provides Text Image Super-Resolution API to improvise the quality and visibility of old and blurred text on an image.

Text Image Super-Resolution can zoom in an image that contains the text and significantly improve the definition of the text.

Introduction

Quality improvement has become crucial in this era of digitalization where all our documents are kept in the folders, shared over the network and read on the digital device.

Imaging the grapple of an elderly person who has no way to read and understand an old prescribed medical document which has gone blurred and deteriorated.

Can we evade such issues??

NO!!

Let’s unbind what Huawei ML Kit offers to overcome such challenges of our day to day life.

Huawei ML Kit provides Text Image Super-Resolution API to improvise the quality and visibility of old and blurred text on an image.

Text Image Super-Resolution can zoom in an image that contains the text and significantly improve the definition of the text.

Limitations

The text image super-resolution service requires images with the maximum resolution 800 x 800 px and the length greater than or equal to 64 px.

Development Overview

Prerequisite

Must have a Huawei Developer Account

Must have Android Studio 3.0 or later

Must have a Huawei phone with HMS Core 4.0.2.300 or later

EMUI 3.0 or later

Software Requirements

Java SDK 1.7 or later

Android 5.0 or later

Preparation

Create an app or project in the Huawei app gallery connect.

Provide the SHA Key and App Package name of the project in App Information Section and enable the ML Kit API.

Download the agconnect-services.json file.

Create an Android project.

Integration

Add below to build.gradle (project)file, under buildscript/repositories and allprojects/repositories.

Maven {url ‘http://developer.huawei.com/repo/'}

Add below to build.gradle (app) file, under dependencies.

To use the Base SDK of ML Kit-Text Image Super Resolution, add the following dependencies:

dependencies{

// Import the base SDK.

implementation ‘com.huawei.hms:ml-computer-vision-textimagesuperresolution:2.0.3.300’

}

To use the Full SDK of ML Kit-Text Image Super Resolution, add the following

dependencies{

// Import the Full SDK.

implementation ‘com.huawei.hms:ml-computer-vision-textimagesuperresolution-model:2.0.3.300’

}

Adding permissions

<uses-permission android:name=”android.permission.CAMERA “ />

<uses-permission android:name=”android.permission.READ_EXTERNAL_STORAGE” />

Automatically Updating the Machine Learning Model

Add the following statements to the AndroidManifest.xml file to automatically install the machine learning model on the user’s device.

<meta-data

android:name=”com.huawei.hms.ml.DEPENDENCY”

android:value= “tisr”/>

Development Process

This article focuses on demonstrating the capabilities of Huawei’s ML Kit: Text Image Super- Resolution API’s.

Here is the example which explains how can we integrate this powerful API to leverage the benefits of improvising the Text-Image quality and provide full accessibility to the reader to read the old and blur newspapers from an online news directory.

TextImageView Activity : Launcher Activity

This is main activity of “The News Express “application.

Activity_text_image_view.xml

This is the view class for the above activity class.

TextImageSuperResolutionActivity

This activity class performs following actions:

Image picker implementation to pick the image from the gallery

Convert selected image to Bitmap

Create a text image super-resolution analyser.

Create an MLFrame object by using android.graphics.Bitmap.

Perform super-resolution processing on the image with text.

Stop the analyser to release detection resources.

activity_text_super_resolution.xml

View file for the above activity.

Results

Conclusion

It’s wonderful to create useful application which provide the user accessibility to elderly with the help of Huawei ML kit.

References

https://github.com/HMS-Core/hms-ml-demo

--

--