… and working mechanism of content-aware feature reassembly for upsampling in several visual
… propose the content-aware reassembly of features (CARAFE) to upsample a feature map. …

… Our CARAFE+ + reassembles the features in a local region by weighted summation. In
summary, spatial attention is a rescaling operator with point-wise guidance while CARAFE++ is …

… Visualization of CARAFE We demonstrate how CARAFE performs content-aware reassembly
… Red units are reassembled into the green center unit by CARAFE in the top-down pathway …

… Contentaware Reassembly of Features (CARAFE) has been proposed to achieve feature …
and even better performance to the original CARAFE, but with much lower complexity, eg, …

… integrates Content-Aware Reassembly of Features (CARAFE) … module to extract effective
features and simplify the network. … CARAFE module for upsampling to avoid the loss of feature …

… Content-aware Reassembly of Features (CARAFE) is a well-designed learnable operation
to achieve feature … and even better performance to the original CARAFE, but with much lower …

… —CARAFE (Content-Aware Reassembly of Features) [32]. Compared with traditional upsampling
methods, CARAFE can … As shown in Figure 13, CARAFE consists of a kernel prediction …

… , the CARAFE (Content-Aware ReAssembly of FEatures) [34] upsampling module is introduced.
The semantic information of feature maps is fully utilized for feature restoration through …

… feature extraction. • … , the CARAFE (Content-Aware ReAssembly of FEatures) [34] upsampling
module is introduced. The semantic information of feature maps is fully utilized for feature …

… The CARAFE module enhances the contextual information of feature maps through
content-aware reassembly, improving feature representation capability. The TripleAttention …