whether the encoders for RGB and depth share weights?

[fereenwong]

Whether the feature extractors for RGB image and depth map share weights? Is the input an RGB-D 4-channel? Or repeat the depth map 3 times?

[Li-Chongyi]

Hi, We used the VGG network as the feature extractor which fixes the weights. We respectively feed the RGB image and the depth map (repeat three times) to this extractor.

fereenwong [email protected] 于2020年12月11日周五 下午3:04写道：

Whether the feature extractors for RGB image and depth map share weights? Is the input an RGB-D 4-channel? Or repeat the depth map 3 times?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/Li-Chongyi/cmMS-ECCV20/issues/2, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIEVQEJGMKQAVMIFJ3C3OR3SUHAAVANCNFSM4UWFGJ5A .

[fereenwong]

Thanks for your reply. Did the fixes the weights means without fine-tuning during the training process?

[Li-Chongyi]

Yes, exactly.

fereenwong [email protected] 于2020年12月11日周五 下午3:45写道：

Thanks for your reply. Did the fixes the weights means without fine-tuning during the training process?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/Li-Chongyi/cmMS-ECCV20/issues/2#issuecomment-743032323, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIEVQEJCXZN2QTT63UXA4GDSUHE2NANCNFSM4UWFGJ5A .

[fereenwong]

Another issue about the ablation study. In the Tab. 2 of the paper, what are the cmFA and cmFC?

[Li-Chongyi]

we noted them in the paper.

发自我的iPhone

在 2020年12月11日，下午4:10，fereenwong [email protected] 写道：

 Another issue about the ablation study. In the Tab. 2 of the paper, what are the cmFA and cmFC?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.