Closed cdzheng2g closed 3 weeks ago
Yes, we used the default coco accuracy metric in mmdet, and the mAP mentioned in the paper is actually mAP50-95:0.05. I suggest you refer to the website for specific information: https://github.com/open-mmlab/mmdetection/blob/main/mmdet/evaluation/metrics/coco_metric.py.
Is the test performance index mAP in the paper experiment mAP50-95?