In S4 E12 of The Big Bang Theory, Leonard comes up with an idea for a smartphone app to identify differential equations. Now I saw this episode recently and have since started working on this app. Currently it just recognizes basic addition, multiplication, subtraction and division equations but I plan to expand it to also include more complex equtions. I also currently have my internship going on so a bit tied up with that as well.
- High Accuracy: Achieves an impressive 97% training accuracy in recognizing handwritten mathematical symbols and equations.
- Image Processing: Utilizes OpenCV for efficient image preprocessing and contour detection.
- Deep Learning Model: Implements a custom resnet based neural network architecture for digit and symbol recognition.
- Equation Parsing: Converts recognized symbols into solvable mathematical expressions.
- Solves equation: The application recognises the equation and solves it.
- CUDA Support: Trained using CUDA for accelerated performance, with CPU inference capabilities for broader accessibility.
- Weights and Biases: Check out the weights and biases report to see how the model training progress went
- Implement more complex mathematical operations recognition
- Enhance the user interface for better interactivity
- Optimize the model for mobile deployment
I've also included a dockerfile for easy deployment. To build and run the docker container:
docker build -t maths-app .
To run it:
docker run -p 8501:8501 maths-app
I'd like to thank wblachowksi for the dataset