《Image Super-Resolution Using Very Deep Residual Channel Attention Networks》阅读笔记
problem
The lowresolution inputs and features contain abundant low-frequency information, which is treated equally across channels, hence hindering the representational ability of CNNs
RCAN
- a residual in residual (RIR) structure : form very deep network;allow abundant low-frequency information to be bypassed through multiple skip connections,making the main network focus on learning high-frequency information.
- a cgannel attention mechanism : adaptively rescale channel-wise features by considering interdependencies among channels
RCAN:G个RG
RG:B个RCAB
RCAB :
。。