Before training a system to process and understand images, it’s important to understand how data is represented:
Two common tasks in computer vision are regression and classification.
At a high-level, the computer system aims use high level feature detection to classify the image into the corresponding class.
One approach is to do manual feature extraction. The problem is that a human is that really good at defining those features. There is a lot of variations that will not be accounted for.
Ideally, we want a system to detect the features automatically.
Reference:
NOTION_PAGE:e43a17b5-0f7e-4b5f-8393-c070b22340a3