Android Tensorflow Digits Recognition

An Android app to recognize YOUR hand-writen character

Contents:

• Overview
• Releases
• Requirements
• CNN Model
• Training Process
• How to use
• Results
• Resources
• Find Me

Overview:

In this project, I will create a simple convolutional neural network [CNN] to classify handwriting digits using MNIST dataset using TensorFlow
The main goal is to deploy this network to android app using TFLite

Releases:

Download The APK file:

• Version 1.0

Requirements:

• Python 3
• Tensorflow v2 (I used v2.1.0)
• Android Studio (I used v3.6.3 with SDK: 29)

CNN Model:

Training Process:

How to use:

Just build the App in android studio, when it runs write whatever you want.

Results:

Some Straightforward inputs:

Used Resources:

• Paint: I used this awesome tutorial https://www.youtube.com/watch?v=uJGcmGXaQ0o
• Visualized results: The famous MPAndroidChart https://github.com/PhilJay/MPAndroidChart