computer vision: yolo and r-cnn
Classification vs localization vs detection
classification is where you label a picture as a class, localisation is where you need to put a bounding box on the object it and detection is detecting one or multiple instances of the class the three together form the basis of many problems like self driving vehicles.
Localization
Along with the classes percentages with localization you get 4 coordinates of the bounding box and whether there’s an object in the picture:
- pc: whether there is a car in the picture
- bx the x axis coordinate of the center of the rectangle
- by the y axis coordinate of the center of the rectangle
- bh the height of the rectangle
- bw the width of the rectangle