Instructions:

1. Download AIDA automatically

2. Read images

3. parse bounding boxes (on symbols)

4. Create DCCNN Architecture

5. Train symbol classifier on handwritten math dataset

6. Predict symbols on AIDA dataset

Dataset Links:

• Math Symbols:
  • https://www.kaggle.com/datasets/xainano/handwrittenmathsymbols/code
• AIDA Dataset:
  • https://www.kaggle.com/datasets/aidapearson/ocr-data
• MNIST Dataset:
  • http://yann.lecun.com/exdb/mnist/
  • https://www.kaggle.com/datasets/hojjatk/mnist-dataset
    • Had to use above kaggle link (other link required login)

Reference Links:

• Towards Data Science Article
  • https://towardsdatascience.com/computer-vision-auto-grading-handwritten-mathematical-answersheets-8974744f72dd
• Code from artical
  • https://github.com/divyaprabha123/Autograding-handwritten-mathematical-worksheets/blob/d074738c2e1db6b4619e9930f576439b309d1f37/utils_functions.py#L636-L725
• Deep Columnar Convoluational Neural Network (DCCNN) Paper
  • https://www.ijcaonline.org/archives/volume145/number12/majumdar-2016-ijca-910772.pdf

K:

• Download AIDA Dataset
• Download MNIST
• Download SYMBOLS
• Invert symbols
• Pad symbols