# Cascade RPN: Delving into High-Quality Region Proposal Network with Adaptive Convolution

arXiv: https://arxiv.org/abs/1909.06720 · HTML: https://mlsanity.com/paper/1909.06720/

## Key facts

- Cited by 57 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 15 Sep 2019.

## Built on by (most cited first, 30 of 57)

- [Probabilistic two-stage detection](https://mlsanity.com/paper/2103.07461/) (2021)
- [Efficient DETR: Improving End-to-End Object Detector with Dense Prior](https://mlsanity.com/paper/2104.01318/) (2021)
- [Learning Open-World Object Proposals without Learning to Classify](https://mlsanity.com/paper/2108.06753/) (2021)
- [Anchor-free Oriented Proposal Generator for Object Detection](https://mlsanity.com/paper/2110.01931/) (2021)
- [Detecting Everything in the Open World: Towards Universal Object Detection](https://mlsanity.com/paper/2303.11749/) (2023)
- [RFLA: Gaussian Receptive Field based Label Assignment for Tiny Object Detection](https://mlsanity.com/paper/2208.08738/) (2022)
- [RefineMask: Towards High-Quality Instance Segmentation with Fine-Grained Features](https://mlsanity.com/paper/2104.08569/) (2021)
- [Abductive Ego-View Accident Video Understanding for Safe Driving Perception](https://mlsanity.com/paper/2403.00436/) (2024)
- [Disentangle Your Dense Object Detector](https://mlsanity.com/paper/2107.02963/) (2021)
- [Semantic Grouping Network for Video Captioning](https://mlsanity.com/paper/2102.00831/) (2021)
- [A New Dataset and A Baseline Model for Breast Lesion Detection in Ultrasound Videos](https://mlsanity.com/paper/2207.00141/) (2022)
- [Small Object Detection via Coarse-to-fine Proposal Generation and Imitation Learning](https://mlsanity.com/paper/2308.09534/) (2023)
- [HS-FPN: High Frequency and Spatial Perception FPN for Tiny Object Detection](https://mlsanity.com/paper/2412.10116/) (2024)
- [Structured Co-reference Graph Attention for Video-grounded Dialogue](https://mlsanity.com/paper/2103.13361/) (2021)
- [TNCR: Table Net Detection and Classification Dataset](https://mlsanity.com/paper/2106.15322/) (2021)
- [Cross-view Masked Diffusion Transformers for Person Image Synthesis](https://mlsanity.com/paper/2402.01516/) (2024)
- [Boosting R-CNN: Reweighting R-CNN Samples by RPN's Error for Underwater Object Detection](https://mlsanity.com/paper/2206.13728/) (2022)
- [Text-based Person Search in Full Images via Semantic-Driven Proposal Generation](https://mlsanity.com/paper/2109.12965/) (2021)
- [Extending One-Stage Detection with Open-World Proposals](https://mlsanity.com/paper/2201.02302/) (2022)
- [Visible and Clear: Finding Tiny Objects in Difference Map](https://mlsanity.com/paper/2405.11276/) (2024)
- [MDSGen: Fast and Efficient Masked Diffusion Temporal-Aware Transformers for Open-Domain Sound Generation](https://mlsanity.com/paper/2410.02130/) (2024)
- [PRVQL: Progressive Knowledge-guided Refinement for Robust Egocentric Visual Query Localization](https://mlsanity.com/paper/2502.07707/) (2025)
- [1st Place Solution for the UVO Challenge on Image-based Open-World Segmentation 2021](https://mlsanity.com/paper/2110.10239/) (2021)
- [Spatio-Temporal Learnable Proposals for End-to-End Video Object Detection](https://mlsanity.com/paper/2210.02368/) (2022)
- [LAD: A Hybrid Deep Learning System for Benign Paroxysmal Positional Vertigo Disorders Diagnostic](https://mlsanity.com/paper/2210.08282/) (2022)
- [Bridging Images and Videos: A Simple Learning Framework for Large Vocabulary Video Object Detection](https://mlsanity.com/paper/2212.10147/) (2022)
- [Edge-guided Representation Learning for Underwater Object Detection](https://mlsanity.com/paper/2306.00440/) (2023)
- [Recursively Refined R-CNN: Instance Segmentation with Self-RoI Rebalancing](https://mlsanity.com/paper/2104.01329/) (2021)
- [PDNet: Toward Better One-Stage Object Detection With Prediction Decoupling](https://mlsanity.com/paper/2104.13876/) (2021)
- [Universal Object Detection with Large Vision Model](https://mlsanity.com/paper/2212.09408/) (2022)

## Abstract (arXiv)

This paper considers an architecture referred to as Cascade Region Proposal Network (Cascade RPN) for improving the region-proposal quality and detection performance by textitsystematically addressing the limitation of the conventional RPN that textitheuristically defines the anchors and textitaligns the features to the anchors. First, instead of using multiple anchors with predefined scales and aspect ratios, Cascade RPN relies on a textitsingle anchor per location and performs multi-stage refinement. Each stage is progressively more stringent in defining positive samples by starting out with an anchor-free metric followed by anchor-based metrics in the ensuing stages. Second, to attain alignment between the features and the anchors throughout the stages, textitadaptive convolution is proposed that takes the anchors in addition to the image features as its input and learns the sampled features guided by the anchors. A simple implementation of a two-stage Cascade RPN achieves AR 13.4 points higher than that of the conventional RPN, surpassing any existing region proposal methods. When adopting to Fast R-CNN and Faster R-CNN, Cascade RPN can improve the detection mAP by 3.1 and 3.5 points, respectively. The code is made publicly available at urlhttps://github.com/thangvubk/Cascade-RPN.git.