Thank you very much for making your code available and congrats on your incredible paper.
I have a question: It seems like you call self.Non_local(eda_out) module first, which leads to the embedded_gaussian operation. Is this not the Residual Nonlocal Temporal Attention Module? RNTA in your paper, fig 2. (e)?
And only after the CALayer1D is called, which seems to be what you describe in the paper as signal channel attention module Fig. 2 (d)?
Dear author,
Thank you very much for making your code available and congrats on your incredible paper.
I have a question: It seems like you call self.Non_local(eda_out) module first, which leads to the embedded_gaussian operation. Is this not the Residual Nonlocal Temporal Attention Module? RNTA in your paper, fig 2. (e)?
And only after the CALayer1D is called, which seems to be what you describe in the paper as signal channel attention module Fig. 2 (d)?
Could you better describe the order?
Thank you.