search

jiyanggao / TALL

TALL: Temporal Activity Localization via Language Query
184 stars 48 forks source link

About the detailed extraction process of C3D features #10

Open zzxihuanheixiu opened 2 years ago

zzxihuanheixiu commented 2 years ago

I read the C3D paper and found that the author used image scaling and clipping when training C3D. I just want to ask you, how do you use the pre-trained C3D model to extract video features? Do I need to do special preprocessing before I input the video clip into the pre-trained C3D model?

If it is convenient, can you provide the code for extracting C3D features?

stha-prashant commented 2 years ago

@zzxihuanheixiu I am looking for the same answer. Have you found a solution yet?

zzxihuanheixiu commented 2 years ago

no,i am sorry

------------------ 原始邮件 ------------------ 发件人: "Prashant @.>; 发送时间: 2022年8月25日(星期四) 中午1:26 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [jiyanggao/TALL] About the detailed extraction process of C3D features (Issue #10)

@zzxihuanheixiu I am looking for the same answer. Have you found a solution yet?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>