search

syuqings / video-paragraph

Codes for paper "Towards Diverse Paragraph Captioning for Untrimmed Videos". CVPR 2021
MIT License
64 stars 3 forks source link

val set of charades dataset #8

Closed ZHONGHZONG closed 3 years ago

ZHONGHZONG commented 3 years ago

Dear professor syuqings,

Thank you very much for your article,Sorry to bother you,Why is the val set of charades dataset the same as the test set? My assessment score was the same as the training model score.

I will be appreciated for your reply as soon as possible, thank you very much!

Best regards,

syuqings commented 3 years ago

Hi, the official split of Charades Captions dataset [Video captioning via hierarchical reinforcement learning] is 6963 for training, 500 for validation and 1760 for test. However, the downloaded annotations have 7473 items labeled as "train", and 1760 labeled as "test". And there is no video labeled as "validation". It seems that the training set and validation set have been merged. Therefore, we just validate the model on the 1760 testing set.

ZHONGHZONG commented 3 years ago

Hi，Thank you very much for your reply,Sorry to be a bother,Happy every day！