论文阅读笔记(四十四):Deconvolutional Networks

Building robust low and mid-level image representations, beyond edge primitives, is a long-standing goal in vision. Many existing feature detectors spatially pool edge information which destroys cues
相关文章
相关标签/搜索