From Features to Classical Machine Learning
In this session, students move beyond basic image manipulations to the extraction of informative features and their use in classical machine learning models.
The focus will be on feature descriptors such as Histograms of Oriented Gradients (HOG), Scale-Invariant Feature Transform (SIFT), Speeded-Up Robust Features (SURF), and Oriented FAST and Rotated BRIEF (ORB). These techniques provide compact, informative vector representations that capture essential visual properties of images, enabling conventional machine learning algorithms to operate effectively.
Students will learn how to translate these features into input vectors suitable for classifiers such as Support Vector Machines and Random Forests. Dimensionality reduction techniques, including Principal Component Analysis (PCA) and t-distributed Stochastic Neighbor Embedding (t-SNE), will be used for visualization and to address the high dimensionality typical of image data.
Practical use cases, such as facial recognition with engineered features or document image analysis, will be explored, highlighting the strengths and limitations of classical approaches.
Required Reading and Listening
Textbooks:
- Hands-On Image Processing with Python By Sandipan Dey, Packt Publishing, November 2018. (GitHub)
- Mastering OpenCV 4 with Python By Alberto Fernández Villán, Packt Publishing, March 2019. (GitHub)
- Python Image Processing Cookbook By Sandipan Dey, Packt Publishing, April 2020. (GitHub)
- Feature Extraction and Image Processing for Computer Vision, 4th Edition By Mark Nixon, Alberto Aguado, Academic Press, November 2019.
