Mask R-CNN

Authors: He, K., Gkioxari, G., Dollár, P., & Girshick, R. Year: 2017 Venue: ICCV ArXiv: https://arxiv.org/abs/1703.06870

Summary

Mask R-CNN extends Faster R-CNN for instance segmentation by adding a branch for predicting object masks. It achieves state-of-the-art on COCO and becomes the standard for instance-level understanding.

Key Concepts

• Instance segmentation framework
• Mask prediction branch
• RoI Align for accurate alignment
• Multi-task learning (detection + segmentation)
• Pixel-level predictions

Impact

Mask R-CNN became the go-to architecture for instance segmentation and influenced many subsequent work on dense predictions. It's still widely used in production systems.

Related Papers

• Faster R-CNN: Region Proposal Networks
• R-CNN: Rich Feature Hierarchies
• Panoptic Segmentation

---

Citation:

@inproceedings{he2017mask,
  title={Mask R-CNN},
  author={He, Kaiming and Gkioxari, Georgia and Dollár, Piotr and Girshick, Ross},
  booktitle={IEEE International Conference on Computer Vision},
  year={2017}
}