https://github.com/rcode879/handwritten-digit-detector

A school machine learning project based on the Knearest neighbours model
https://github.com/rcode879/handwritten-digit-detector

csv-files knearest-neighbor-algorithm knearest-neighbor-classification knearest-neighbor-classifier machine-learning opencv python3 tkinter tkinter-gui

Last synced: 17 days ago
JSON representation

A school machine learning project based on the Knearest neighbours model

• Host: GitHub
• URL: https://github.com/rcode879/handwritten-digit-detector
• Owner: Rcode879
• Created: 2024-01-29T19:47:13.000Z (11 months ago)
• Default Branch: main
• Last Pushed: 2024-01-30T13:22:16.000Z (11 months ago)
• Last Synced: 2024-01-30T22:47:49.670Z (11 months ago)
• Topics: csv-files, knearest-neighbor-algorithm, knearest-neighbor-classification, knearest-neighbor-classifier, machine-learning, opencv, python3, tkinter, tkinter-gui
• Language: Python
• Homepage:
• Size: 2.66 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Handwritten-digit-detector

A python program that trains a Knearest neighbours model on the MNIST data set(which is in the form of a CSV file) to recognise handwritten digits.

Implemented into a TKinter GUI which had a button to train the model, a button to load in your own png file with a handwritten digit and a button to output the model accuracy.

## Required packages: 

-CSV 

-Tkinter 

-OpenCV 

-Scikit.learn

## Warning:

- line 16 with variable named "mnist_train.csv", is the csv file file containing the mnist data set to train the model, the file was too large to upload to the repository

- line 96 is the background image of the gui - "gui.png" - you can add your own peronal image by downloading one of your choice and renaming it

- If you wish to test your own image(png), it must be 28 by 28 pixels with a black background and the number written in white

- This model is NOT the most accurate and would most likely be more accurate in a neural network so do not expect perfect predicitons