VisDrone
commented
3 years ago font{
line-height: 1.6;
}
ul,ol{
padding-left: 20px;
list-style-position: inside;
}
Thank you for the suggestion. We will improve the evaluation process of VisDrone and modify the leaderboard following the popular benchmark datasets like COCO.
360037320
***@***.***
签名由
网易邮箱大师
定制
On 5/7/2021 ***@***.***> wrote： Hi, I'm not sure if this project is still maintained. When I was studying on vehicle detection and tracking with VisDrone, there's seldom researches that I can compete with. I think the main resean is that there's no public annotations for test-challenge subset and most of the teams didn't report their local evaluation results on test-dev. So, like COCO, maybe it's more preferable for organizers to recommend participates to report results of both test-dev and test-chanllege (of course only the latter one is taken in to consideration for competition).
—You are receiving this because you are subscribed to this thread.Reply to this email directly, view it on GitHub, or unsubscribe.
Hi, I'm not sure if this project is still being maintained. When I was studying on vehicle detection and tracking with VisDrone, there's seldom researches that I can compete with. I think the main reason is that there's no public annotations for test-challenge subset and most of the teams didn't report their local evaluation results on test-dev.
So, like COCO, maybe it's more preferable for organizers to recommend participates to report results of both test-dev and test-chanllege (of course only the latter one is taken in to consideration for competition).