HPL123
Missing something
The unseen classifier is not been published, or how to generate unseen classes classifier.
The unseen classifier is not been published, or how to generate unseen classes classifier.
Could anyone please tell me that the gzsd results generated when I use VOC are quite different from the original author's results. What needs to be modified when implementing GZSD?
I met the same problem, how to solve it?
Do you solve it? When you implemented GZSD on the VOC data set, were the results consistent with the author's? When I implemented it, I only had half of the APs. Is there anything that needs to be modified?
Could anyone please tell me that the gzsd results generated when I use VOC are quite different from the original author's results. What needs to be modified when implementing GZSD?
I meet the same problem, did you solve it? My ZSD experiment can only achieve 5 points less than the mAP in papar even used it's published "classifier_best.pth".
I'm sorry that I don't have the problem you meet. There are still some other questions I haven't overcome.
------------------ 原始邮件 ------------------ 发件人: "HPL123/RRFS" @.>; 发送时间: 2024年1月29日(星期一) 下午3:09 @.>; @.@.>; 主题: Re: [HPL123/RRFS] Missing something (Issue #1)
Could anyone please tell me that the gzsd results generated when I use VOC are quite different from the original author's results. What needs to be modified when implementing GZSD?
I meet the same problem, did you solve it? My ZSD experiment can only achieve 5 points less than the mAP in papar even used it's published "classifier_best.pth".
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>