YOLO V3 anchor boxes and map calculation #8518
Unanswered
ahmad-omran
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
in YOLOv3: An Incremental Improvement, Joseph Redmon, Ali Farhadi mintion that they used COCO dataset and the 9 clusters were:
(10,13); (16,30); (33,23); (30,61); (62,45); (59,119); (116, 90); (156,198); (373,326).
and map= 55.3 for YOLOv3-416
i used 'get_coco_dataset.sh' in script file to download coco dataset and used the command
'!./darknet detector calc_anchors cfg/coco.data -num_of_clusters 9 -width 416 -height 416 -show'
the result was 'anchors = 13, 19, 32, 47, 47,114, 99, 55, 134,129, 80,226, 308,145, 175,285, 351,339' and avg IoU = 56.97 %
when i used the command
'!./darknet detector map cfg/obj.data cfg/yolov3.cfg yolov3.weights'
the result was 'mean average precision (mAP@0.50) = 0.546563, or 54.66 % '
Are my results correct?
Beta Was this translation helpful? Give feedback.
All reactions