/Multiple-Handwritten-Digit-Recognition

Multiple Handwritten Digit Recognition app Using Deep Learing - CNN from Canvas build on tkinter- GUI

Primary LanguagePython

Multiple Handwritten Digit Recognition using Deep Learning ( TensorFlow - Keras )

Requirements

  • TensorFlow (Keras)
  • Python 3.5 +
  • Numpy (+ mkl for Windows)
  • PIL ( pillow )
  • Opencv
  • tkinter ( python GUI )

About Project

  • Trained model on MNIST dataset Using CNN (Convolutional Nueral Network)

  • Save model as 'mnist.h5' ( train_digit_recognizer.py )

  • Using tkinter GUI make a canvas and write digit on it

  • Using PIL get a copy of Handwritten Digits on the canvas and saved into '/img' as 'img_{image_number}.png'

  • Also with the help of OpenCV by recognizing contours it can handle multiple digits

  • Use saved model 'mnist.h5' to predict saved Handwritten Digit image from canvas

Screenshots

  • Drawing canvas ...

  • Output Imgae ...

  • Saved Image using PIL-ImageGrab ...