image2num

• 

  website interface

Inspiration

We were inspired by educational math web applications, and wanted to implement machine learning somehow into a similar program.

What it does

Recognizes numbers from an input image using neural network training

How we built it

PyTorch -> MNIST handwritten digits dataset

Challenges we ran into

Neural network training: Transformers, Loading and using the trained model with model.load_state_dict

Accomplishments that we're proud of

Getting the model to work + implementation

What we learned

Machine learning/NN training, PyTorch, Figma + CSS UI editing, venv + Flask web development

What's next for image2num

recognizing math equations, and converting into LaTeX mathematical notation setting up a website host, currently running on local host

slideshow: https://docs.google.com/presentation/d/11sdbez8Fs6PzhY9WM6QirZcVJzz0CEAj7fJdHzkvf1k/edit?usp=sharing

Built With

• css
• figma
• flask
• git
• github
• html
• html5
• mnist
• python
• pytorch
• venv