# Dynamic Sparse R-CNN

arXiv: https://arxiv.org/abs/2205.02101 · HTML: https://mlsanity.com/paper/2205.02101/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- Builds on 27 of its 35 references that are in the map.
- First posted on arXiv 4 May 2022.

## Builds on (most cited first, 27 of 27)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://mlsanity.com/paper/1409.1556/) (2014)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks](https://mlsanity.com/paper/1506.01497/) (2015)
- [End-to-End Object Detection with Transformers](https://mlsanity.com/paper/2005.12872/) (2020)
- [Rethinking the Inception Architecture for Computer Vision](https://mlsanity.com/paper/1512.00567/) (2015)
- [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://mlsanity.com/paper/2010.04159/) (2020)
- [SSD: Single Shot MultiBox Detector](https://mlsanity.com/paper/1512.02325/) (2015)
- [Fast R-CNN](https://mlsanity.com/paper/1504.08083/) (2015)
- [YOLOv3: An Incremental Improvement](https://mlsanity.com/paper/1804.02767/) (2018)
- [FCOS: Fully Convolutional One-Stage Object Detection](https://mlsanity.com/paper/1904.01355/) (2019)
- [YOLO9000: Better, Faster, Stronger](https://mlsanity.com/paper/1612.08242/) (2016)
- [Cascade R-CNN: Delving into High Quality Object Detection](https://mlsanity.com/paper/1712.00726/) (2017)
- [Objects as Points](https://mlsanity.com/paper/1904.07850/) (2019)
- [Focal Loss for Dense Object Detection](https://mlsanity.com/paper/1708.02002/) (2017)
- [Bridging the Gap Between Anchor-based and Anchor-free Detection via Adaptive Training Sample Selection](https://mlsanity.com/paper/1912.02424/) (2019)
- [Sparse R-CNN: End-to-End Object Detection with Learnable Proposals](https://mlsanity.com/paper/2011.12450/) (2020)
- [Conditional DETR for Fast Training Convergence](https://mlsanity.com/paper/2108.06152/) (2021)
- [R-FCN: Object Detection via Region-based Fully Convolutional Networks](https://mlsanity.com/paper/1605.06409/) (2016)
- [Dynamic Convolution: Attention over Convolution Kernels](https://mlsanity.com/paper/1912.03458/) (2019)
- [CondConv: Conditionally Parameterized Convolutions for Efficient Inference](https://mlsanity.com/paper/1904.04971/) (2019)
- [OTA: Optimal Transport Assignment for Object Detection](https://mlsanity.com/paper/2103.14259/) (2021)
- [Rethinking Transformer-based Set Prediction for Object Detection](https://mlsanity.com/paper/2011.10881/) (2020)
- [Probabilistic Anchor Assignment with IoU Prediction for Object Detection](https://mlsanity.com/paper/2007.08103/) (2020)
- [DyNet: Dynamic Convolution for Accelerating Convolutional Neural Networks](https://mlsanity.com/paper/2004.10694/) (2020)
- [Light-Head R-CNN: In Defense of Two-Stage Object Detector](https://mlsanity.com/paper/1711.07264/) (2017)

## Built on by (most cited first, 11 of 11)

- [FAQ: Feature Aggregated Queries for Transformer-based Video Object Detectors](https://mlsanity.com/paper/2303.08319/) (2023)
- [Groupwise Query Specialization and Quality-Aware Multi-Assignment for Transformer-based Visual Relationship Detection](https://mlsanity.com/paper/2403.17709/) (2024)
- [A Review of Transformer-Based Models for Computer Vision Tasks: Capturing Global Context and Spatial Relationships](https://mlsanity.com/paper/2408.15178/) (2024)
- [Anchor3DLane++: 3D Lane Detection via Sample-Adaptive Sparse 3D Anchor Regression](https://mlsanity.com/paper/2412.16889/) (2024)
- [Table Detection for Visually Rich Document Images](https://mlsanity.com/paper/2305.19181/) (2023)
- [ASAG: Building Strong One-Decoder-Layer Sparse Detectors via Adaptive Sparse Anchor Generation](https://mlsanity.com/paper/2308.09242/) (2023)
- [Rethinking Detection Based Table Structure Recognition for Visually Rich Document Images](https://mlsanity.com/paper/2312.00699/) (2023)
- [Beyond Boxes: Mask-Guided Spatio-Temporal Feature Aggregation for Video Object Detection](https://mlsanity.com/paper/2412.04915/) (2024)
- [FlowDet: Unifying Object Detection and Generative Transport Flows](https://mlsanity.com/paper/2512.16771/) (2025)
- [Can the Query-based Object Detector Be Designed with Fewer Stages?](https://mlsanity.com/paper/2309.16306/) (2023)
- [RBF Weighted Hyper-Involution for RGB-D Object Detection](https://mlsanity.com/paper/2310.00342/) (2023)

## Abstract (arXiv)

Sparse R-CNN is a recent strong object detection baseline by set prediction on sparse, learnable proposal boxes and proposal features. In this work, we propose to improve Sparse R-CNN with two dynamic designs. First, Sparse R-CNN adopts a one-to-one label assignment scheme, where the Hungarian algorithm is applied to match only one positive sample for each ground truth. Such one-to-one assignment may not be optimal for the matching between the learned proposal boxes and ground truths. To address this problem, we propose dynamic label assignment (DLA) based on the optimal transport algorithm to assign increasing positive samples in the iterative training stages of Sparse R-CNN. We constrain the matching to be gradually looser in the sequential stages as the later stage produces the refined proposals with improved precision. Second, the learned proposal boxes and features remain fixed for different images in the inference process of Sparse R-CNN. Motivated by dynamic convolution, we propose dynamic proposal generation (DPG) to assemble multiple proposal experts dynamically for providing better initial proposal boxes and features for the consecutive training stages. DPG thereby can derive sample-dependent proposal boxes and features for inference. Experiments demonstrate that our method, named Dynamic Sparse R-CNN, can boost the strong Sparse R-CNN baseline with different backbones for object detection. Particularly, Dynamic Sparse R-CNN reaches the state-of-the-art 47.2% AP on the COCO 2017 validation set, surpassing Sparse R-CNN by 2.2% AP with the same ResNet-50 backbone.