Detect hand-writing digits and symbols (draw by mouse). Solve the math operations and recommend some related lectures
1/ Dataset: The dataset on Kaggle is here. I filter it and choose only 21000 images for my dataset: (0-9) x y sqrt * / - + < >
2/ Model: I use CNN model to classify them.
3/ Preprocess:
image = tf.image.decode_jpeg(image, channels=3)
image = tf.image.rgb_to_grayscale(image)
image = tf.image.resize(image, [28, 28])
image = (255 - image)/255.0
-
Input image from dataset:
-
After Preprocess function, the image is like this:
4/ Some symbol functions work on this web app:
| Operation | Example |
|---|---|
| + - * / | a + b |
| Sqrt | sqrt(a^2 + b^2) |
| Equation 1 | ax+b |
| Equation 2 | ax^2+bx+c=0 |
| Equation 3 | ax^3 + bx^2 + cx + d=0 |
| Equation overlaps | ax^4 + bx^2 + c=0 |
| Inequation 1 | ax+b > 0 or ax+b < 0 |
| Inequation 2 | ax^2 + bx + c > 0 or ax^2 + bx + c < 0 |
| Inequation 3 | ax^3 + bx^2 + cx+d > 0 or ax^3 + bx^2 + cx + d < 0 |
| Graph equation 1 | y=ax+b |
| Graph equation 2 | y=ax^2+bx+c=0 |
| Graph equation 3 | y=ax^3 + bx^2 + cx + d=0 |
| Acreage & Perimeter of Square | a = m; S=m^2; P = 4m |
| Acreage & Perimeter of Rectangle | a = m; b = n; S=m^n; P = 2(m+n) |
Note:
-
Sqrt must be drew first.
-
If you don't draw "=", it will solve the equation with default "=0"
5/ Processing after model predicting.
- I find some labels which confuse the model prediction.
- Find special pixels to classify them again.
6/ Solve the result.
Calculate the math function depending on type of math and show the result.
7/ Recommend some lectures related to the type of math operation.
Depending on result of prediction, web app will classify the type of math and recommend some lectures related to it.
HO HUY HOA Email: hoahh2201@gmail.com