Doodle Vision is a mini-project aimed to replicate Google's Quick, Draw Game. Project Consists of 2 parts, training a model capable of classifying various doodles, and the web application . The web application allows users to draw a prompt, which is then sent to the deep learning model for prediction. The model returns the top 5 guesses, and the application displays whether the model correctly identified the drawing.
- Drawing Canvas: Allows users to draw sketches based on a given prompt using the
react-sketch-canvaslibrary. - Real-time Predictions: The model provides live feedback after each stroke.
- Prediction Feedback: Displays the top 5 guesses from the model.
- Frontend: React,
react-sketch-canvas - Backend: Python, Flask
- Machine Learning: PyTorch, ResNet-18 architecture
The classifier utilises ResNet-18 trained on 23 classes with 3000 images per class. After 10 epochs (due to limitied hardware capacity),the model achieved a validation accuracy of 88.15%.
The dataset used for training the model is sourced from Kaggle's Doodle Dataset.
- Start the React frontend:
cd src/ npm start - Start the backend server (for model predictions):
cd backend/ python prediction_model_server.py
- Utilize entire dataset during training phase.
- Experiment with deeper archtectures.
- Set up live guesses, akin to Quick! Draw by Google