RetinaNet

Introduced by Lin et al. in Focal Loss for Dense Object Detection

RetinaNet is a one-stage object detection model that utilizes a focal loss function to address class imbalance during training. Focal loss applies a modulating term to the cross entropy loss in order to focus learning on hard negative examples. RetinaNet is a single, unified network composed of a backbone network and two task-specific subnetworks. The backbone is responsible for computing a convolutional feature map over an entire input image and is an off-the-self convolutional network. The first subnet performs convolutional object classification on the backbone's output; the second subnet performs convolutional bounding box regression. The two subnetworks feature a simple design that the authors propose specifically for one-stage, dense detection.

We can see the motivation for focal loss by comparing with two-stage object detectors. Here class imbalance is addressed by a two-stage cascade and sampling heuristics. The proposal stage (e.g., Selective Search, EdgeBoxes, DeepMask, RPN) rapidly narrows down the number of candidate object locations to a small number (e.g., 1-2k), filtering out most background samples. In the second classification stage, sampling heuristics, such as a fixed foreground-to-background ratio, or online hard example mining (OHEM), are performed to maintain a manageable balance between foreground and background.

In contrast, a one-stage detector must process a much larger set of candidate object locations regularly sampled across an image. To tackle this, RetinaNet uses a focal loss function, a dynamically scaled cross entropy loss, where the scaling factor decays to zero as confidence in the correct class increases. Intuitively, this scaling factor can automatically down-weight the contribution of easy examples during training and rapidly focus the model on hard examples.

Formally, the Focal Loss adds a factor $(1 - p_{t})^\gamma$ to the standard cross entropy criterion. Setting $\gamma>0$ reduces the relative loss for well-classified examples ($p_{t}>.5$), putting more focus on hard, misclassified examples. Here there is tunable focusing parameter $\gamma \ge 0$.

$$ {\text{FL}(p_{t}) = - (1 - p_{t})^\gamma \log\left(p_{t}\right)} $$

Source: Focal Loss for Dense Object Detection

Latest Papers

PAPER DATE
Globally-scalable Automated Target Recognition (GATR)
Gary ChernAusten GroenerMichael HarnerTyler KuhnsAndy LamStephen O'NeillMark Pritt
2020-09-10
What leads to generalization of object proposals?
Rui WangDhruv MahajanVignesh Ramanathan
2020-08-13
Deep Learning-based Human Detection for UAVs with Optical and Infrared Cameras: System and Experiments
Timo HinzmannTobias StegemannCesar CadenaRoland Siegwart
2020-08-10
Synthetic to Real Unsupervised Domain Adaptation for Single-Stage Artwork Recognition in Cultural Sites
Giovanni PasqualinoAntonino FurnariGiovanni SignorelloGiovanni Maria Farinella
2020-08-04
Multi-Class 3D Object Detection Within Volumetric 3D Computed Tomography Baggage Security Screening Imagery
Qian WangNeelanjan BhowmikToby P. Breckon
2020-08-03
Probabilistic Anchor Assignment with IoU Prediction for Object Detection
| Kang KimHee Seok Lee
2020-07-16
AQD: Towards Accurate Quantized Object Detection
Jing LiuBohan ZhuangPeng ChenMingkui TanChunhua Shen
2020-07-14
A Systematic Evaluation of Object Detection Networks for Scientific Plots
Pritha GangulyNitesh MethaniMitesh M. KhapraPratyush Kumar
2020-07-05
Working with scale: 2nd place solution to Product Detection in Densely Packed Scenes [Technical Report]
| Artem Kozlov
2020-06-14
Concurrent Segmentation and Object Detection CNNs for Aircraft Detection and Identification in Satellite Images
Damien GrosgeorgeMaxime ArbelotAlex GoupilleauTugdual CeillierRenaud Allioux
2020-05-27
Introduction of a new Dataset and Method for Detecting and Counting the Pistachios based on Deep Learning
| Mohammad RahimzadehAbolfazl Attar
2020-05-08
RetinaMask: A Face Mask detector
Mingjie JiangXinqi FanHong Yan
2020-05-08
Scale-Equalizing Pyramid Convolution for Object Detection
| Xinjiang WangShilong ZhangZhuoran YuLitong FengWayne Zhang
2020-05-06
Global Table Extractor (GTE): A Framework for Joint Table Identification and Cell Structure Recognition Using Visual Context
Xinyi ZhengDoug BurdickLucian PopaNancy Xin Ru Wang
2020-05-01
Learning Gaussian Maps for Dense Object Detection
Sonaal Kant
2020-04-24
Tracking by Instance Detection: A Meta-Learning Approach
Guangting WangChong LuoXiaoyan SunZhiwei XiongWenjun Zeng
2020-04-02
RetinaTrack: Online Single Stage Joint Detection and Tracking
Zhichao LuVivek RathodRonny VotelJonathan Huang
2020-03-30
On the Evaluation of Prohibited Item Classification and Detection in Volumetric 3D Computed Tomography Baggage Security Screening Imagery
Qian WangNeelanjan BhowmikToby P. Breckon
2020-03-27
Multi-Plateau Ensemble for Endoscopic Artefact Segmentation and Detection
| Suyog JadhavUdbhav BambaArnav ChavanRishabh TiwariAryan Raj
2020-03-23
PointINS: Point-based Instance Segmentation
Lu QiXiangyu ZhangYingcong ChenYukang ChenJian SunJiaya Jia
2020-03-13
DEEVA: A Deep Learning and IoT Based Computer Vision System to Address Safety and Security of Production Sites in Energy Industry
Nimish M. AwalgaonkarHaining ZhengChristopher S. Gurciullo
2020-03-02
Sperm Detection and Tracking in Phase-Contrast Microscopy Image Sequences using Deep Learning and Modified CSR-DCF
| Mohammad reza MohammadiMohammad RahimzadehAbolfazl Attar
2020-02-11
A learning without forgetting approach to incorporate artifact knowledge in polyp localization tasks
Roger D. Soberanis-MukulMaxime KayserAnna-Maria ZverevaPeter KlareNassir NavabShadi Albarqouni
2020-02-07
Single-Stage Object Detection from Top-View Grid Maps on Custom Sensor Setups
Sascha WirgesShuxiao DingChristoph Stiller
2020-02-03
Comparison of object detection methods for crop damage assessment using deep learning
Ali HamidiSepehrSeyed Vahid MirnezamiJason K. Ward
2019-12-31
Benchmark for Generic Product Detection: A Low Data Baseline for Dense Object Detection
| Srikrishna VaradarajanSonaal KantMuktabh Mayank Srivastava
2019-12-19
The Benefits of Close-Domain Fine-Tuning for Table Detection in Document Images
| Ángela Casado-GarcíaCésar DomínguezJónathan HerasEloy MataVico Pascual
2019-12-12
AugFPN: Improving Multi-scale Feature Learning for Object Detection
| Chaoxu GuoBin FanQian ZhangShiming XiangChunhong Pan
2019-12-11
Learning from Noisy Anchors for One-stage Object Detection
Hengduo LiZuxuan WuChen ZhuCaiming XiongRichard SocherLarry S. Davis
2019-12-11
SpineNet: Learning Scale-Permuted Backbone for Recognition and Localization
| Xianzhi DuTsung-Yi LinPengchong JinGolnaz GhiasiMingxing TanYin CuiQuoc V. LeXiaodan Song
2019-12-10
Side-Aware Boundary Localization for More Precise Object Detection
Jiaqi WangWenwei ZhangYuhang CaoKai ChenJiangmiao PangTao GongJianping ShiChen Change LoyDahua Lin
2019-12-09
Bridging the Gap Between Anchor-based and Anchor-free Detection via Adaptive Training Sample Selection
| Shifeng ZhangCheng ChiYongqiang YaoZhen LeiStan Z. Li
2019-12-05
Multiple Anchor Learning for Visual Object Detection
| Wei KeTianliang ZhangZeyi HuangQixiang YeJianzhuang LiuDong Huang
2019-12-04
GhostNet: More Features from Cheap Operations
| Kai HanYunhe WangQi TianJianyuan GuoChunjing XuChang Xu
2019-11-27
Filter Response Normalization Layer: Eliminating Batch Dependence in the Training of Deep Neural Networks
| Saurabh SinghShankar Krishnan
2019-11-21
Evaluating the Transferability and Adversarial Discrimination of Convolutional Neural Networks for Threat Object Detection and Classification within X-Ray Security Imagery
Yona Falinie A. GausNeelanjan BhowmikSamet AkcayToby P. Breckon
2019-11-20
Traffic Sign Detection and Recognition for Autonomous Driving in Virtual Simulation Environment
Meixin ZhuJingyun HuZiyuan PuZhiyong CuiLiangwu YanYinhai Wang
2019-10-27
ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks
| Qilong WangBanggu WuPengfei ZhuPeihua LiWangmeng ZuoQinghua Hu
2019-10-08
RandAugment: Practical automated data augmentation with a reduced search space
| Ekin D. CubukBarret ZophJonathon ShlensQuoc V. Le
2019-09-30
ASSD: Attentive Single Shot Multibox Detector
| Jingru YiPengxiang WuDimitris N. Metaxas
2019-09-27
WiderPerson: A Diverse Dataset for Dense Pedestrian Detection in the Wild
Shifeng ZhangYiliang XieJun WanHansheng XiaStan Z. LiGuodong Guo
2019-09-25
Synthetic dataset generation for object-to-model deep learning in industrial applications
| Matthew Z. WongKiyohito KuniiMax BaylisWai Hong OngPavel KroupaSwen Koller
2019-09-24
Road Damage Detection Acquisition System based on Deep Neural Networks for Physical Asset Management
A. A. AnguloJ. A. Vega-FernándezL. M. Aguilar-LoboS. NatrajG Ochoa-Ruiz
2019-09-19
STELA: A Real-Time Scene Text Detector with Learned Anchor
| Linjie DengYanxiang GongXinchen LuYi LinZheng MaMei Xie
2019-09-17
Aerial multi-object tracking by detection using deep association networks
Ajit JadhavPrerana MukherjeeVinay KaushikBrejesh Lall
2019-09-04
ScarfNet: Multi-scale Features with Deeply Fused and Redistributed Semantics for Enhanced Object Detection
Jin Hyeok YooDongsuk KumJun Won Choi
2019-08-01
DR Loss: Improving Object Detection by Distributional Ranking
| Qi QianLei ChenHao LiRong Jin
2019-07-23
Semi-supervised Breast Lesion Detection in Ultrasound Video Based on Temporal Coherence
Sihong ChenWeiping YuKai MaXinlong SunXiaona LinDesheng SunYefeng Zheng
2019-07-16
Cascade RetinaNet: Maintaining Consistency for Single-Stage Object Detection
Hongkai ZhangHong ChangBingpeng MaShiguang ShanXilin Chen
2019-07-16
Learning Data Augmentation Strategies for Object Detection
| Barret ZophEkin D. CubukGolnaz GhiasiTsung-Yi LinJonathon ShlensQuoc V. Le
2019-06-26
Pattern Spotting in Historical Documents Using Convolutional Models
Ignacio ÚbedaJose M. SaavedraStéphane NicolasCaroline PetitjeanLaurent Heutte
2019-06-20
Stand-Alone Self-Attention in Vision Models
| Prajit RamachandranNiki ParmarAshish VaswaniIrwan BelloAnselm LevskayaJonathon Shlens
2019-06-13
NAS-FCOS: Fast Neural Architecture Search for Object Detection
| Ning WangYang GaoHao ChenPeng WangZhi TianChunhua ShenYanning Zhang
2019-06-11
Improving RetinaNet for CT Lesion Detection with Dense Masks from Weak RECIST Labels
| Martin ZlochaQi DouBen Glocker
2019-06-05
Fully Quantized Network for Object Detection
Rundong Li Yan Wang Feng Liang Hongwei Qin Junjie Yan Rui Fan
2019-06-01
Light-Weight RetinaNet for Object Detection
| Yixing LiFengbo Ren
2019-05-24
Spatial Group-wise Enhance: Improving Semantic Feature Learning in Convolutional Networks
| Xiang LiXiaolin HuJian Yang
2019-05-23
Accurate Face Detection for High Performance
Faen ZhangXinyu FanGuo AiJianfei SongYongqiang QinJiahong Wu
2019-05-05
Attention Augmented Convolutional Networks
| Irwan BelloBarret ZophAshish VaswaniJonathon ShlensQuoc V. Le
2019-04-22
NAS-FPN: Learning Scalable Feature Pyramid Architecture for Object Detection
| Golnaz GhiasiTsung-Yi LinRuoming PangQuoc V. Le
2019-04-16
Single Pixel Reconstruction for One-stage Instance Segmentation
Jun YuJinghan YaoJian ZhangZhou YuDacheng Tao
2019-04-16
Evaluation of a Dual Convolutional Neural Network Architecture for Object-wise Anomaly Detection in Cluttered X-ray Security Imagery
Yona Falinie A. GausNeelanjan BhowmikSamet AkçayPaolo M. Guillen-GarciaJack W. BarkerToby P. Breckon
2019-04-10
Libra R-CNN: Towards Balanced Learning for Object Detection
| Jiangmiao PangKai ChenJianping ShiHuajun FengWanli OuyangDahua Lin
2019-04-04
FCOS: Fully Convolutional One-Stage Object Detection
| Zhi TianChunhua ShenHao ChenTong He
2019-04-02
DetNAS: Backbone Search for Object Detection
| Yukang ChenTong YangXiangyu ZhangGaofeng MengXinyu XiaoJian Sun
2019-03-26
Feature Selective Anchor-Free Module for Single-Shot Object Detection
| Chenchen ZhuYihui HeMarios Savvides
2019-03-02
Towards Pedestrian Detection Using RetinaNet in ECCV 2018 Wider Pedestrian Detection Challenge
Md Ashraful Alam Milton
2019-02-04
Consistent Optimization for Single-Shot Object Detection
Tao KongFuchun SunHuaping LiuYuning JiangJianbo Shi
2019-01-19
Region Proposal by Guided Anchoring
| Jiaqi WangKai ChenShuo YangChen Change LoyDahua Lin
2019-01-10
RetinaMask: Learning to predict masks improves state-of-the-art single-shot detection for free
| Cheng-Yang FuMykhailo ShvetsAlexander C. Berg
2019-01-10
AutoFocus: Efficient Multi-Scale Inference
| Mahyar NajibiBharat SinghLarry S. Davis
2018-12-04
Deep Regionlets: Blended Representation and Deep Learning for Generic Object Detection
Hongyu XuXutao LvXiaoyu WangZhou RenNavaneeth BodlaRama Chellappa
2018-11-28
M2Det: A Single-Shot Object Detector based on Multi-Level Feature Pyramid Network
| Qijie ZhaoTao ShengYongtao WangZhi TangYing ChenLing CaiHaibin Ling
2018-11-12
Biologically-plausible learning algorithms can scale to large datasets
Will XiaoHonglin ChenQianli LiaoTomaso Poggio
2018-11-08
DropBlock: A regularization method for convolutional networks
| Golnaz GhiasiTsung-Yi LinQuoc V. Le
2018-10-30
Salience Biased Loss for Object Detection in Aerial Images
Peng SunGuang ChenGuerdan LukeYi Shang
2018-10-18
DetNet: Design Backbone for Object Detection
Zeming LiChao PengGang YuXiangyu ZhangYangdong DengJian Sun
2018-09-01
Fast and Accurate, Convolutional Neural Network Based Approach for Object Detection from UAV
Xiaoliang WangPeng ChengXinchuan LiuBenedict Uzochukwu
2018-08-16
MetaAnchor: Learning to Detect Objects with Customized Anchors
Tong YangXiangyu ZhangZeming LiWenqiang ZhangJian Sun
2018-07-03
YOLOv3: An Incremental Improvement
| Joseph RedmonAli Farhadi
2018-04-08
Group Normalization
| Yuxin WuKaiming He
2018-03-22
Focal Loss Dense Detector for Vehicle Surveillance
Xiaoliang WangPeng ChengXinchuan LiuBenedict Uzochukwu
2018-03-03
MegDet: A Large Mini-Batch Object Detector
| Chao PengTete XiaoZeming LiYuning JiangXiangyu ZhangKai JiaGang YuJian Sun
2017-11-20
Focal Loss for Dense Object Detection
| Tsung-Yi LinPriya GoyalRoss GirshickKaiming HePiotr Dollár
2017-08-07

Categories