iSketch

Android application for recognizing simple hand-drawn images.

The application uses a nine layerd CNN for classification and runs instantly on any android phone. It was trained on a subset of Quick Draw Dataset with an input shape of 28*28. Currently it supports 100 classes (objects) and can be easily trained within few minutes, using a standard GPU !!!

Prerequisites

• Android Studio
• Android Smartphone

Features

• A paint canvas with draw & erase.
• Recognizes 100 different objects/shapes.
• Shows top three predictions.
• Provides voice output with tts.

Screenshot

Versioning

Version 1.0

Authors

Anil Sathyan

Acknowledgments

• "https://medium.com/tensorflow/train-on-google-colab-and-run-on-the-browser-a-case-study-8a45f9b1474e"
• "https://github.com/tensorflow/examples/tree/master/lite/examples/image_classification/android"
• "https://guides.codepath.com/android/Basic-Painting-with-Views"