[paper reading] C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection CVPR2019
MIL陷入局部最优，检测到局部，无法完整的检测到物体。将instance划分为空间相关和类别相关的子集。在这些子集中定义一系列平滑的损失近似代替原损失函数，优化这些平滑损失。
C-MIL learns instance subsets, where the instances are spatially related, i.e., overlapping with each other, and class related, i.e., having similar object class scores.
SRE实战 互联网时代守护先锋，助力企业售后服务体系运筹帷幄！一键直达领取阿里云限量特价优惠。![[paper reading] C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection CVPR2019 随笔 第1张](https://img.liuyixiang.com/image/2019/4/UVzENb.png "[paper reading] C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection CVPR2019 随笔 第1张")
C-MIL treats images as bags and image regions generated by an object proposal method [24,32] as instances
![[paper reading] C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection CVPR2019 随笔 第2张](https://img.liuyixiang.com/image/2019/4/IFnMFb.png "[paper reading] C-MIL: Continuation Multiple Instance Learning for Weakly Supervised Object Detection CVPR2019 随笔 第2张")
更多精彩
