From YOLO you should not only get the detections but also the classes. Your model was most likely trained on the COCO dataset which has a certain table of objects that it can offer.
You can find such a list here: https://github.com/pjreddie/darknet/blob/master/data/coco.names
You can see that the class "person" is on the 0'th position on the list, so if you only use the detections from that class, you will only get the bounding boxes with persons in them!
Yes, I have also done the same method.But the person class detects the humans who are walking on the road .But I need only the Pedestrian who are using zebra crossing to be detected.So for that is there any special algorithm for pedestrian alone,