{
  "id": 241470,
  "title": "Taking User's Handwritten Digit and Predicting Output w/ Accuracy [OpenCV, Tkinter, Keras] | 99.28% Accuracy Model",
  "url": "/competitions/digit-recognizer/discussion/241470",
  "author_name": "Arya Shah",
  "post_date": "2021-05-24T17:03:24.668000",
  "votes": 2,
  "comment_count": 0,
  "views": null,
  "content": "<p>I tried building a simple model and documenting the same for the purpose of building an interactive desktop application using the Tkinter Module in Python.</p>\n<p>Apart from predictions using test data, A novel way to predict how well the model performed would be to make use of user's handwritten input which was made possible using OpenCV and Tkinter Module to build the GUI.<br>\nThe user can interact with the drawing canvas by using the mouse to draw the desired digit (multiple digits can be drawn next to each other). The predict button then places the digits in individual bounding boxes and predicts what digit it is with the accuracy for the same.</p>\n<p>**Complete well documented Github Repository with Code and Theory: <a href=\"https://github.com/aryashah2k/Handwritten-Multiple-Digits-Recognizer**\" target=\"_blank\">https://github.com/aryashah2k/Handwritten-Multiple-Digits-Recognizer**</a></p>\n<p><em>The notebook containing the Training and saving the model can be found on my profile.</em></p>\n<p>Do suggest what more could be done when we have a model available with us… Text to speech of the output from the user input seems interesting. Any follow ups or ideas would be highly appreciated. Excited to hear from the discussion community here :)</p>",
  "messages": [
    {
      "id": 1321400,
      "postDate": "2021-05-24T17:03:24.667Z",
      "content": "<p>I tried building a simple model and documenting the same for the purpose of building an interactive desktop application using the Tkinter Module in Python.</p>\n<p>Apart from predictions using test data, A novel way to predict how well the model performed would be to make use of user's handwritten input which was made possible using OpenCV and Tkinter Module to build the GUI.<br>\nThe user can interact with the drawing canvas by using the mouse to draw the desired digit (multiple digits can be drawn next to each other). The predict button then places the digits in individual bounding boxes and predicts what digit it is with the accuracy for the same.</p>\n<p>**Complete well documented Github Repository with Code and Theory: <a href=\"https://github.com/aryashah2k/Handwritten-Multiple-Digits-Recognizer**\" target=\"_blank\">https://github.com/aryashah2k/Handwritten-Multiple-Digits-Recognizer**</a></p>\n<p><em>The notebook containing the Training and saving the model can be found on my profile.</em></p>\n<p>Do suggest what more could be done when we have a model available with us… Text to speech of the output from the user input seems interesting. Any follow ups or ideas would be highly appreciated. Excited to hear from the discussion community here :)</p>",
      "rawMarkdown": "I tried building a simple model and documenting the same for the purpose of building an interactive desktop application using the Tkinter Module in Python.\n\nApart from predictions using test data, A novel way to predict how well the model performed would be to make use of user's handwritten input which was made possible using OpenCV and Tkinter Module to build the GUI.\nThe user can interact with the drawing canvas by using the mouse to draw the desired digit (multiple digits can be drawn next to each other). The predict button then places the digits in individual bounding boxes and predicts what digit it is with the accuracy for the same.\n\n**Complete well documented Github Repository with Code and Theory: https://github.com/aryashah2k/Handwritten-Multiple-Digits-Recognizer**\n\n*The notebook containing the Training and saving the model can be found on my profile.*\n\nDo suggest what more could be done when we have a model available with us... Text to speech of the output from the user input seems interesting. Any follow ups or ideas would be highly appreciated. Excited to hear from the discussion community here :)",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1321400": "I tried building a simple model and documenting the same for the purpose of building an interactive desktop application using the Tkinter Module in Python.\n\nApart from predictions using test data, A novel way to predict how well the model performed would be to make use of user's handwritten input which was made possible using OpenCV and Tkinter Module to build the GUI.\nThe user can interact with the drawing canvas by using the mouse to draw the desired digit (multiple digits can be drawn next to each other). The predict button then places the digits in individual bounding boxes and predicts what digit it is with the accuracy for the same.\n\n**Complete well documented Github Repository with Code and Theory: https://github.com/aryashah2k/Handwritten-Multiple-Digits-Recognizer**\n\n*The notebook containing the Training and saving the model can be found on my profile.*\n\nDo suggest what more could be done when we have a model available with us... Text to speech of the output from the user input seems interesting. Any follow ups or ideas would be highly appreciated. Excited to hear from the discussion community here :)"
  }
}