search

praveena2j / RJCMA

ABAW6 (CVPR-W) We achieved second place in the valence arousal challenge of ABAW6
https://affective-behavior-analysis-in-the-wild.github.io/6th/
9 stars 2 forks source link

Model Implementation Detail Inquiry #3

Open liangyh1519 opened 2 weeks ago

liangyh1519 commented 2 weeks ago

In the implementation of your project, how is the temporal dimension K of the multimodal feature representation collapsed or eliminated after passing through the Multilayer Perceptron (MLP) layer? I have carefully reviewed the associated paper and model code but am still unable to comprehend this process. Is there a dimensionality reduction or temporal pooling technique applied here? If possible, could you provide some specific implementation details or direct me to the correct part of the code? Thank you very much for your assistance.

praveena2j commented 2 weeks ago

In this project we are doing frame level prediction so we are not doing any temporal aggregation in the temporal dimension. Hope it helps.

liangyh1519 commented 2 weeks ago

Thanks for your valuable response!