For each grid cell with an object, the YOLO model predicts the probability value that the object belongs to each preset category. Combined with the confidence score, the final score corresponding to each category of each bounding box can be calculated to determine the specific category of the object and complete the classification link in the detection task.
