Coder School FTMLE in-class exercise based on hoahh2201 project about "OCR Handwritten Digits and Maths Performance". The lite version can detect handwritten sketch on canvas and can do simple plus + and minus - operations. The final evaluated result will be returned.
- tensorflow 2.0
- flask
- imutils
- mahotas
- opencv
- Clone this repository to your local machine.
- Change directory to model_ml, train the
multidigit_classifier.ipynb
withcomplete_data
set. - Fine tune and export the model.
- Work on
preprocess_with_opencv.ipynb
to transform a raw image into what our model can predict. - Finish the discussions at the end of the notebook.
- Move to
app
and copy the code you did onpreprocess_with_opencv.ipynb
tomain.py
regarding the "YOUR PART" pointer.
๐ต Extra:
- Re-design the flaskapp or Adding more maths expression to the model.
- Get a good sleep because if you reach this point, you are probably tired and really in need of some self care ๐๐ปโโ๏ธ Come back later for more challenge. ๐