Handwritten Equation Solver trained by handwritten digits and mathematical symbol using Convolutional Neural Network with some image processing techniques
Equation can contain any digit from 0-9 and symbol +,x,- Works on image with white background and digits/symbols are in black.
You can run all the three ipynb files either separately or sequentially.
- For running Data_extraction.ipynb first download train images.rar zip file and extract it in the folder containing Data_extaction.ipynb file.
- For running model_training.ipynb, you either need to download train_final.csv or you can run it after succesfully running Data_extraction.ipynb.
- For running CNN_test.ipynb, you either need to download model_final.h5 and model_final.json file or you can run it after succesfully running model_training.ipynb file. You also need to replace the path of the image in code from the local path of image to be tested on your computer.
- Input an image containing a handwritten equation. Convert the image to a binary image and then invert the image(if digits/symbols are in black).
- We obtain contours of the image by default, it will obtain contours from left to right.
- Obtain bounding rectangle for each contour.
- Sometimes, we may get two or more contours for the same digit/symbol. To avoid that, we can check if the bounding rectangle of those two contours overlaps or not. If they overlap, then discard the smaller rectangle.
- Now, resize all the remaining bounding rectangle to 28 by 28.
- Using our model, predict the corresponding digit/symbol for each bounding rectangle and store it in a string. 7.After that use ‘eval’ function on the string to solve the equation.
