推荐genericpseudo · 2018-03
“Another thread worth pulling on is Mask R-CNN: Mask R-CNN”
推荐 Mask R-CNN 作为值得深挖的另一个方向
在 Faster R-CNN 上加掩码分支,同时做检测与实例分割
提出用于目标实例分割的 Mask R-CNN:在 Faster R-CNN 已有的边框识别分支之外,并行增加一个预测每个实例掩码的分支,训练简单、额外开销小,约 5 fps。该框架还可推广到人体姿态估计,在 COCO 的实例分割、框检测和关键点检测三项任务上都取得领先结果。
评论者把它作为目标检测和实例分割方向值得跟进的新进展提及,并用它说明某些检测库当时还来不及收录最新检测器。
“Another thread worth pulling on is Mask R-CNN: Mask R-CNN”
“Another detector is the recently released Mask-RCNN [3], which of course wouldn't be possible to be included in this publication as we can't travel through time yet.”