We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
您好, 我在github 上阅读了您的论文关于GLD-v2数据集的图像检索。有个问题想请教一下关于评估。我所下载的数据集GLD-v2中大部分类别都是小于10张的。所以请问map@100这个评估方法你们是在所有的训练数据集中使用的吗?还是只是对于图像超过100张的类别计算的?
很期待您的回复 Tianyi Hu
The text was updated successfully, but these errors were encountered:
map@100是用在测试集的,与训练集无关,只要某个Query(查询图片)有至少1张GT都能算。
例如某Query其GT为M和P,Query在Index(底库,假设有N张图片)中搜索,会获得所有N张Index图片与Query图片的相似度,按相似度从大到小排序即Q: ABCDM...(省略95个,这中间没有P)qaP(这个P排103位)
则map@all = 0.5 * (1/5 + 2/103)
map@100 = 0.5 * (1/5)
在图片检索领域,一些人把map@100错误地算成map@100 = (1/5),认为P不在前100可以不记做分母,参考谷歌的论文和delg代码,谷歌是按正确方法计算的。评价方式更多讨论,可以见https://stackoverflow.com/questions/54966320/mapk-computation
Sorry, something went wrong.
No branches or pull requests
您好,
我在github 上阅读了您的论文关于GLD-v2数据集的图像检索。有个问题想请教一下关于评估。我所下载的数据集GLD-v2中大部分类别都是小于10张的。所以请问map@100这个评估方法你们是在所有的训练数据集中使用的吗?还是只是对于图像超过100张的类别计算的?
很期待您的回复
Tianyi Hu
The text was updated successfully, but these errors were encountered: