Anyone know how to calculate the mAP coco dataset for every class with my own tensorflow object detection model?
