I've been examining the STVEN_rPPGNet, and I noticed that you didn't include the temporal encoder-decoder (ED) structure in your network, and I believe you have the code for ED) in your repo. In your previous work, PhysNet, 3D-CNN with ED structure has the best performance, and I'm curious as to why you decided not to include it in your current project.
I believe that adding the ED structure to your current network could potentially improve its performance, and I would like to understand your thought process behind not including it. Is there a particular reason why you chose not to use the ED structure, or did you simply overlook it?
I would appreciate any insights you can provide, as I'm interested in learning more about your work and the decisions you make in your research.
Dear Yu,
I've been examining the STVEN_rPPGNet, and I noticed that you didn't include the temporal encoder-decoder (ED) structure in your network, and I believe you have the code for ED) in your repo. In your previous work, PhysNet, 3D-CNN with ED structure has the best performance, and I'm curious as to why you decided not to include it in your current project.
I believe that adding the ED structure to your current network could potentially improve its performance, and I would like to understand your thought process behind not including it. Is there a particular reason why you chose not to use the ED structure, or did you simply overlook it?
I would appreciate any insights you can provide, as I'm interested in learning more about your work and the decisions you make in your research.
Thank you for your time and consideration.
Sincerely, Chen