Project 3c - Image classification #3

Project start: 23 December 2021

Project finish: 31 December 2021

Jetson Nano setup to date: 194 hours

Jetson Nano programming to date: 112 hours

As a beginner, I found the setup quite difficult.

Objective: a non-Udemy example of convolutional neural networks

This project is for classification of circles, rectangles and triangles. Data is for 100 .png files of each shape and is downloaded from Kaggle.

I queried the trained neural network with three images that I drew on pieces of paper with a Sharpie and took pictures of with my Samsung S20. The neural network correctly predicted the type of each.

The following .zip file contains a Jupyter Notebook program based on a program from the Udemy course (to run Jupyter Notebook, type jupyter-notebook):

project_3c_V6.ipynb.zip

Download ZIP • 39KB

The following .zip file contains the required data:

data_project_3c.zip

Download ZIP • 440KB

The image below shows the loss over 200 epochs.

The image below shows the accuracy over 200 epochs.

If you have any problems or need clarification please contact me: jscanova@gmail.com