Paper Reading: Recombinator Networks: Learning Coarse-to-Fine Feature Aggregation （CVPR 2016）

Github 源碼： https://github.com/SinaHonari/RCN convnet 存在的問題： max-pooling: for tasks requiring precise localization, such as pixel level prediction and segmentation,max-pooling destroys exactly the info

>>阅读原文<<