Hands-On Image Processing with Python

Classical versus deep learning

• Handcrafted versus automated feature extraction: In order to solve image processing problems with traditional ML techniques, the most important preprocessing step is the handcrafted feature (for example, HOG and SIFT) extraction in order to reduce the complexity of an image and make patterns more visible for learning algorithms to work. The biggest advantage of deep learning algorithms is that they try to learn low-level and high-level features from training images in an incremental manner. This eliminates the need for handcrafted feature in extraction or engineering.
• By parts versus end-to-end solution: Traditional ML techniques solve the problem statement by breaking down the problem, solving different parts ...