Can I get the same result with the paper?

[limbee]

Hi. Thank you for sharing your implementation.

Is it possible to get the same (or similar) mAP with the paper?
I would be really appreciated if you could tell me the final performance.

[wuguangbin1230]

@limbee what the performance did you get?

[lxtGH]

@wuguangbin1230 @limbee what the performance did you get ?

[XuSen123]

@limbee @wuguangbin1230 @lxtGH what the performance did you get?

[FoolishMao]

@limbee @wuguangbin1230 @lxtGH @XuSen123 what the performance did you get?

[triangleCZH]

@limbee @wuguangbin1230 @lxtGH @XuSen123 @FoolishMao what the performance did you get?

[ghost]

@limbee @wuguangbin1230 @lxtGH @XuSen123 @FoolishMao @triangleCZH what the performance did you get?

[wangkangnian]

@limbee @wuguangbin1230 @lxtGH @XuSen123 @FoolishMao @triangleCZH @zhanglanqing what the performance did you get?

[XipengLin]

@limbee @wuguangbin1230 @lxtGH @XuSen123 @FoolishMao @triangleCZH @zhanglanqing what the performance did you get?

[ZhangGongjie]

Can't u just post your performance? (￣_￣|||)

[jeong-tae]

Hello you guys. I heard that Retina performance on VOC may 55~60mAP which is very lower than SSD, YOLO variations. Maybe... because average number of boxes on images?

I hope this helps you guys.