HandWrittenDigit-Recognition-With-Tkinter-GUI

This repository contains Hand written Digit Recognition Prject with a GUI made using Tkinter.

DEMO

https://youtu.be/pSpwV16Z4_Y

https://www.instagram.com/p/CCP5XQ3j7FS/?utm_source=ig_web_copy_link

Requirements

--->joblib
--->sklearn
--->numpy
--->skimage
--->collections
--->opencv
--->PIL
--->Tkinter

This consits of 2 python files
1.classifiergenerator.py
2.recognizer.py

classifergenerator.py

This file contains python code for generating a classifier using sklearn's Linear SVC model.
We will use mnist dataset to tarin the model.

Step by step flow:

1.loading mnist dataset.
2.Extracting hog features as numpy array.
3.training the model.
4.Saving the model. --->This will generate a classifier model (pickle file) classifier.pkl

recognizer.py

This file contains python code which allows user to draw the digits using mouse pointer and recognize the digits.

Step by step flow:

1.Loading the classifier model
2.Creating GUI(3 buttons:save,predict,clear and canvas to draw)
3.Saving the sketch as image and loading saved image
4.Convertig imge into grayscale
5.Applying gaussianblur
6.Thresholding image
7.Finding contours
8.Drawing rectangle ariund the contours
9.Recognizing the digit
10.Putting text(number) on top of the rectangle

Running

Open commandprompt and change direcrtory to project nad run'python recognizer.py'.
this will open a GUI which consists of 3 buttons.
1.save
2.predict
3.clear

Steps for realtime recognition
step1 - draw a digit
step2 - click save
step3 - click predict--->this will show a popup image with the recognised value written on the sketch.
step4 - clear--->this will clear the canvas

step1-./ .
.
.
.