Skip to content

This project helps exercising my machine learning modelling skills and evaluation techniques on image-based models.

Notifications You must be signed in to change notification settings

kmock930/Machine-Learning-for-Images

Repository files navigation

Image-Based Machine Learning

This project aims to exercise my machine learning modelling and evaluation techniques using tools including Jupyter, Scikit-learn and other ML tools.

Dataset

Salzburg Texture Image Database of color texture images

  • It is a subset of database with images in .pnm format.
  • Images are in three texture categories - Fabric, Gravel, Wood.

Implementation Steps

  1. Load the images into numpy matrices.
  2. Read and Preprocess the images, by applying processes like the histogram equalization.
  3. Visualize a sample image for easier debugging.
  4. Implement a binary logistic regression classifier to distinguish grayscale images of the categories.
  5. Evaluate and Generalize the model accordingly.
  6. Perform Cross-Validation on the testing set.
  7. Implement a multi-class logistic regression model to classify all grayscale images into multiple (3) classes.
  8. Evaluate and Generalize the model accordingly.