Closed MiriamFarber closed 3 years ago
You can refer to the bottom up home page to extract the bottom up feature for your own dataset huang-yq17 邮箱huang-yq17@mails.tsinghua.edu.cn 签名由 网易邮箱大师 定制 On 02/25/2021 21:17, Miriam wrote: I have my own data set which I'd like to train on end-to-end image captioning. Looking at the data/README file, I see that bottom-up features are required. Is it possible to train without having these features? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.
I have my own data set which I'd like to train on end-to-end image captioning. Looking at the data/README file, I see that bottom-up features are required. Is it possible to train without having these features?