search

Finspire13 / CMCS-Temporal-Action-Localization

Completeness Modeling and Context Separation for Weakly Supervised Temporal Action Localization (CVPR2019)
MIT License
150 stars 17 forks source link

Regarding your paper #5

Closed lianxxx closed 5 years ago

lianxxx commented 5 years ago

Thank you first for your code.

In your ablation studies, there is a result for "No TA". During training, how was the "MIL loss" calculated in this case? (Is there an MIL loss for each branch or only one in total for the averaged of all branches or else?)

Finspire13 commented 5 years ago

"No TA" means the temporal attention is replaced with global average/max pooling (See here).

Other configurations are the same.