What is yolo algorithm?

Yolo is a target detection algorithm.

the task of target detection is to find out the objects from the picture and give their categories and positions. for a single picture, the output is the center position (x,y), width (w), height (h) and their categories of each of the n objects contained in the picture.

Yolo's prediction is based on the whole picture, and all the detected target signals, including their categories and positions, are output at one time. Yolo first divides the picture into sxs grid of the same size.

introduction

Yolo only requires that the center of the object identified in the grid must be within this grid (specifically, if the center point of a target is within a grid, the probability of the grid outputting the target category is 1, and the prediction probability of all other grids is set to ).

implementation method: let sxs boxes each predict b bounding boxes, and the bounding boxes have five quantities, namely x,y,h,w of the object and the confidence of prediction; Each grid predicts b bounding box and object categories, and the categories are represented by one-hot.