Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Authors: Ren, S., He, K., Girshick, R., & Sun, J. Year: 2015 Venue: NeurIPS ArXiv: https://arxiv.org/abs/1506.01497

Summary

Faster R-CNN introduces Region Proposal Networks (RPN) to replace selective search. It's significantly faster than R-CNN while maintaining accuracy, becoming the standard for two-stage detection.

Key Concepts

• Region Proposal Networks (RPN)
• Anchor boxes for proposals
• Shared convolutional features
• End-to-end training
• Efficient detection pipeline

Impact

Faster R-CNN dominated detection benchmarks for years and inspired Mask R-CNN. RPN remains fundamental to modern detection.

Related Papers

• R-CNN: Rich Feature Hierarchies
• Mask R-CNN: Instance Segmentation
• YOLO: Real-Time Detection

---

Citation:

@inproceedings{ren2015faster,
  title={Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks},
  author={Ren, Shaoqing and He, Kaiming and Girshick, Ross and Sun, Jian},
  booktitle={Advances in Neural Information Processing Systems},
  year={2015}
}