Skip to content

This network classifies hand-drawn doodles. Trained on the QuickDraw dataset (by Google) consisting of 345 classes and over 100k images per class. Deployed on Android along with a RESTful API for Cloud Computing

Notifications You must be signed in to change notification settings

ananyaroy1011/ScribbleNet-master

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ScribbleNet - Play Store (Android)

Download the Android App here: https://play.google.com/store/apps/details?id=doodle.classifier

ScribbleNet

The notebook includes downloading the classes.txt file as well as the entire dataset of 'Quick Draw' images

The 100_classes.txt consists of 100 common classes (a subset of the 345 original classes)

The number of images used per class is limited to 16000 to increase training speed

With my architecture I achieved a Validation Accuracy of 96%

Instructions to train the model:

Run the Doodle.ipynb on Google Colab (GPU for more disk memory)

Instructions for Flask:

In the command prompt:

python run.py

Now, Navigate to 'localhost:5000' on your browser

Android App:

Cup:

Cup

Envelope:

Envelope

T-shirt:

Tshirt

Flask App:

Suitcase:

Suitcase

Headphones:

Headphones

Lightning:

Lightning

About

This network classifies hand-drawn doodles. Trained on the QuickDraw dataset (by Google) consisting of 345 classes and over 100k images per class. Deployed on Android along with a RESTful API for Cloud Computing

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published