論文閱讀筆記(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
相關文章
相關標籤/搜索