Modeling for Computer Vision
Computer vision tasks are among the most popular problems in practical applications of machine learning; they were the gateway into deep learning for many Kagglers, including yours truly (Konrad, that is). Over the last few years, there has been tremendous progress in the field and new SOTA libraries continue to be released. In this chapter, we will give you an overview of the most popular competition types in computer vision:
- Image classification
- Object detection
- Image segmentation
We will begin with a short section on image augmentation, a group of task-agnostic techniques that can be applied to different problems to increase the generalization capability of our models.