A WebApp for Solve Handwritten Equations using Convolutional Neural Networks
It works with any digit from 0-9 and the symbols +,-,x
White blackground and black digits are necessary
- Two Convolutional Layers
- Two MaxPooling Layers
- Dropout Layer
- Flatten Layer
- Two Dense Layers using Relu
- Output Layer using Softmax
- Adam Optimizer
- Create a virtual environment:
python3 -m venv venv-
Activate the virtual environment:
- On Linux
. venv/bin/activate- On Windows
venv\Scripts\activate
-
Install the dependencies
pip install -r requirements.txt
In the folder model, there is a pre-build model of the neural network and the weights of that model. But if you wanna build or train in your own hands, you need to create a folder called 'datasets' in the root of the project, download the datasets from here and extract the folders '0, 1, 2, 3, 4 ,5, 6, 7, 8, 9, +, -, times (rename the folder for x)' in the folder
After that you are ready to start the application. For that is just:
python -m flask run --without-threads- TensorFlow / Keras
- Python / Flask
- OpenCv
- P5.js