what is AR@AN results of rescaled feature on THUMOS14?

[xinyeCH]

Hi, Tianwei Lin, Thank you for your contribution! I have do some experiments on thumos14 based on your training code for ActivityNet! During feature encoding, I rescale the feature sequence to 100, but I got poor results that AR@100 is about 22%. I notice that you do not rescale feature on Thumos14.
Have you do experiments based on rescaled feature on Thumos14 and what about the results? Does the operation of rescaling have so much impact on the results?
Thank you very much!

[TianweiLin]

Hi @yinianjimo, since the duration of thumos14 videos are very long, the rescale procedure is not suitable for THUMOS14 dataset (while the duration of ActivityNet videos are limited). I have not tried to use rescaled feature on THUMOS, and I suggest that sliding window should be adopted.

[Jockey]

Hi @yinianjimo, since the duration of thumos14 videos are very long, the rescale procedure is not suitable for THUMOS14 dataset (while the duration of ActivityNet videos are limited). I have not tried to use rescaled feature on THUMOS, and I suggest that sliding window should be adopted.

Hi @wzmsltw ， i want to reproduce the experiments on THUMOS14，could you release some details about the setting on THUMOS？

[TianweiLin]

attached. Jockey <[e-mail]> 于2020年5月29日周五 下午10:17写道：

…

Hi @yinianjimo <https://github.com/yinianjimo>, since the duration of thumos14 videos are very long, the rescale procedure is not suitable for THUMOS14 dataset (while the duration of ActivityNet videos are limited). I have not tried to use rescaled feature on THUMOS, and I suggest that sliding window should be adopted. Hi @wzmsltw <https://github.com/wzmsltw> ， i want to reproduce the experiments on THUMOS14，could you release some details about the setting on THUMOS？ — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#30 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AD5SIBUYUJ3L4HLRD47BCK3RT67XVANCNFSM4HCV6FDA> .

[Jockey]

attached. Jockey [e-mail] 于2020年5月29日周五 下午10:17写道：
…
Hi @yinianjimo https://github.com/yinianjimo, since the duration of thumos14 videos are very long, the rescale procedure is not suitable for THUMOS14 dataset (while the duration of ActivityNet videos are limited). I have not tried to use rescaled feature on THUMOS, and I suggest that sliding window should be adopted. Hi @wzmsltw https://github.com/wzmsltw ， i want to reproduce the experiments on THUMOS14，could you release some details about the setting on THUMOS？ — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#30 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AD5SIBUYUJ3L4HLRD47BCK3RT67XVANCNFSM4HCV6FDA .

Thanks a lot! Here is my email : [e-mail]
By the way, could you send me the generated proposals on THUMOS14?