Mask R-CNN[1] is a framework for object instance segmentation, which adds a branch for predicting an object mask in parallel with the existing branch for bounding box recognition of Faster R-CNN. There are some details of reading and implementing it.

Contents

• Paper & Code & note
• Paper understanding
• Code understanding
• Note
• References

Paper & Code & note

Paper: Mask R-CNN(ICCV 2017 paper)
Code: Pytorch
Note: Mendeley

Paper

Abstract

1. It is a framework for object instance segmentation.
2. It extends Faster R-CNN by adding a branch for predicting an object mask in parallel with the existing branch for bounding box recognition.
3. It challenges instance segmentation, bounding-box object detection, and person keypoint detection.
4. It serves as a solid baseline in instance-level recognition.

Problem Description

Problem Solution

Conceptual Understanding

Core Conception

Loss

Mask

RoIAlign

Experiments

Code

The complete code can be found in detectron2[2].

Note

More details of Mask R-CNN and its extends like RoIAlign, bilinear interpolation and etc. can be found in [3].

References

[1] He K, Gkioxari G, Dollár P, et al. Mask r-cnn[C]//Proceedings of the IEEE international conference on computer vision. 2017: 2961-2969.
[2] detectron2. https://github.com/facebookresearch/detectron2
[3] stone. “The amazing Mask R-CNN.” https://zhuanlan.zhihu.com/p/37998710