Methods > Computer Vision > Object Detection Models

Cascade R-CNN

Introduced by Cai et al. in Cascade R-CNN: Delving into High Quality Object Detection

Cascade R-CNN is an object detection architecture that seeks to address problems with degrading performance with increased IoU thresholds (due to overfitting during training and inference-time mismatch between IoUs for which detector is optimal and the inputs). It is a multi-stage extension of the R-CNN, where detector stages deeper into the cascade are sequentially more selective against close false positives. The cascade of R-CNN stages are trained sequentially, using the output of one stage to train the next. This is motivated by the observation that the output IoU of a regressor is almost invariably better than the input IoU.

Cascade R-CNN does not aim to mine hard negatives. Instead, by adjusting bounding boxes, each stage aims to find a good set of close false positives for training the next stage. When operating in this manner, a sequence of detectors adapted to increasingly higher IoUs can beat the overfitting problem, and thus be effectively trained. At inference, the same cascade procedure is applied. The progressively improved hypotheses are better matched to the increasing detector quality at each stage.

Source: Cascade R-CNN: Delving into High Quality Object Detection

Latest Papers

PAPER DATE
OPANAS: One-Shot Path Aggregation Network Architecture Search for Object Detection
| TingTing LiangYongtao WangZhi TangGuosheng HuHaibin Ling
2021-03-08
Augmenting Proposals by the Detector Itself
Xiaopei WanZhenhua GuoChao HeYujiu YangFangbo Tao
2021-01-28
SyNet: An Ensemble Network for Object Detection in UAV Images
| Berat Mert AlbabaSedat Ozer
2020-12-23
Hierarchical Context Embedding for Region-based Object Detection
Zhao-Min ChenXin JinBorui ZhaoXiu-Shen WeiYanwen Guo
2020-08-04
A Solution to Product detection in Densely Packed Scenes
| Tianze RongYanjia ZhuHongxiang CaiYichao Xiong
2020-07-23
Seeing without Looking: Contextual Rescoring of Object Detections for AP Maximization
| Lourenco V. Pato Renato Negrinho Pedro M. Q. Aguiar
2020-06-01
PBRnet: Pyramidal Bounding Box Refinement to Improve Object Localization Accuracy
Li XiaoYufan LuoChunlong LuoLianhe ZhaoQuanshui FuGuoqing YangAnpeng HuangYi Zhao
2020-03-10
Side-Aware Boundary Localization for More Precise Object Detection
| Jiaqi WangWenwei ZhangYuhang CaoKai ChenJiangmiao PangTao GongJianping ShiChen Change LoyDahua Lin
2019-12-09
IMMVP: An Efficient Daytime and Nighttime On-Road Object Detector
Cheng-En WuYi-Ming ChanChien-Hung ChenWen-Cheng ChenChu-Song Chen
2019-10-15
CBNet: A Novel Composite Backbone Network Architecture for Object Detection
| Yudong LiuYongtao WangSiwei WangTing-Ting LiangQijie ZhaoZhi TangHaibin Ling
2019-09-09
InstaBoost: Boosting Instance Segmentation via Probability Map Guided Copy-Pasting
| Hao-Shu FangJianhua SunRunzhong WangMinghao GouYong-Lu LiCewu Lu
2019-08-21
Deep High-Resolution Representation Learning for Visual Recognition
| Jingdong WangKe SunTianheng ChengBorui JiangChaorui DengYang ZhaoDong LiuYadong MuMingkui TanXinggang WangWenyu LiuBin Xiao
2019-08-20
Rethinking Classification and Localization for Cascade R-CNN
Ang LiXue YangChongyang Zhang
2019-07-27
Cascade R-CNN: High Quality Object Detection and Instance Segmentation
| Zhaowei CaiNuno Vasconcelos
2019-06-24
Spatial Group-wise Enhance: Improving Semantic Feature Learning in Convolutional Networks
| Xiang LiXiaolin HuJian Yang
2019-05-23
GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond
| Yue CaoJiarui XuStephen LinFangyun WeiHan Hu
2019-04-25
Hybrid Task Cascade for Instance Segmentation
| Kai ChenJiangmiao PangJiaqi WangYu XiongXiaoxiao LiShuyang SunWansen FengZiwei LiuJianping ShiWanli OuyangChen Change LoyDahua Lin
2019-01-22
Acquisition of Localization Confidence for Accurate Object Detection
| Borui JiangRuixuan LuoJiayuan MaoTete XiaoYuning Jiang
2018-07-30
Cascade R-CNN: Delving into High Quality Object Detection
| Zhaowei CaiNuno Vasconcelos
2017-12-03

Tasks

TASK PAPERS SHARE
Object Detection 14 70.00%
Classification 2 10.00%
Dense Object Detection 1 5.00%
Object Localization 1 5.00%
Pose Estimation 1 5.00%
Object Recognition 1 5.00%

Components

COMPONENT TYPE
RoIAlign
RoI Feature Extractors (optional)
RPN
Region Proposal (optional)

Categories