Download presentation
Presentation is loading. Please wait.
1
rlj
2
Motivation Instance segmentation Proposal based Proposal free 慢
回归算法，每个pixel回归到center的距离
3
基于回归的方法 缺点： Inference的时候需要先得到中心点（聚类算法），再根据距离划 分instance
4
Framework
5
Learnable margin e： 像素 C：中心点 缺点： Margin fixed
6
Learnable margin Since for each instance k the gaussian outputs a foreground background probability map, this can be optimized by using a binary classification loss with the binary foreground background map of each instance as ground-truth. As opposed to using the standard cross-entropy loss function, we opt for using the Lovasz-hinge loss instead.
7
Seed map
8
Post processing Seed map取最大值，得到中心点 在同一个位置，取σ，每个像素预测到中心点距离
根据公式 ，>0.5则属于该中心点所属 instance 从seed map上把整个instance mask掉，取下一个最大值
9
Other details
10
Experiments
11
Experiments
Similar presentations
