This project is a web application for recognizing handwritten digits using machine learning. It provides a simple interface where users can train, test the model. It also stores the model weights so that the next time, they can be loaded directly. Just, upload a image of digit (dims:28x28) and get predictions from the trained model.
Application requires Python3 to run effectively. If you don't have Python installed, you can download it from here.
You can also use Docker to deploy it in your local.
After you finished installing Python, you can install by following the steps below:
git clone https://github.com/Vishesh-dd4723/digit-recogniser-web-app.git
cd digit-recogniser-web-app
# Without Docker
# Install Virtual environment library
pip install virtualenv
# Creating a seperate environment with name "myenv"
python -m venv myenv
# Activating the virtual env
myenv\Scripts\activate
# Install requirements
pip install -r requirements.txt
# Run the main.py file
python "./app/main.py"
#With Docker
# Building an image
docker build -t <image-name> .
# Deploying the image
docker run --rm --name <container-name> -d -p 8080:8080 <image-name>- After running the
main.py, the server will start. - Open
http://localhost:8080/docsin your browser - You will see the swagger UI of the supported endpoints and schemas.
- Initialize the model by hitting
POST http://localhost:8080/digitRecogniser/initialize - Then you can choose the file to be used from training or simply load (if weights exist).
- Then you can use any image to predict.
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change. Some of the things to be worked on:
- At the moment, the app only works on pre-present
csvfiles for training. So create a mechanism to upload and perform CRUD operations on files. - The app consists of only back-end. Front-end is need to be created.