目标检测数据集研究综述

doi:10.11871/jfdc.issn.2096-742X.2024.02.016

摘要/Abstract

摘要：

【应用背景】目标检测是计算机视觉的基本研究问题之一，目标检测数据集是评估目标检测方法性能的基础。【目的】对目标检测领域发展过程中产生的数据集进行分析和介绍可以有效地揭示目标检测数据集的特点、发展趋势以及检测研究面临的主要问题，从时间和领域的角度展现目标检测数据集的现状，一定程度上也可以为研究人员提供数据集使用参考。【方法】主要从目标检测领域通用数据集和包含行人检测、人脸检测、交通道路场景目标检测、航空遥感检测、文本检测多个应用场景的特定领域数据集两个角度出发，关注数据集的挑战性，列举分析应用最为广泛且具有差异的数据集，给出不同场景数据集的图像示例并分析其主要挑战。【结论】对目标检测领域数据集进行介绍的同时，也揭示了目标检测数据集的重要意义、不同场景下的挑战性和特点以及构建目标检测数据集的主要挑战与未来发展趋势。

关键词: 目标检测, 数据集, 行人检测, 人脸检测, 计算机视觉

Abstract:

[Application Background] Object detection is one of the basic research issues in computer vision, and the object detection dataset is the basis for evaluating the performance of object detection methods. [Purpose] Analyzing and introducing the datasets generated with the development of the object detection field can effectively reveal the characteristics, development trends, and main problems faced in object detection research. It can also present the status of object detection datasets from the perspective of time and domain, and to some extent, provide reference for researchers to use datasets. [Method] From the perspectives of the general datasets of object detection and the specific datasets that include multiple application scenarios such as pedestrian detection, face detection, traffic and road scene object detection, aviation remote sensing detection, and text detection, we focus on the challenges of datasets, list and analyze the most widely used and diverse datasets, provide image examples of different scene datasets, and analyze their main challenges. [Conclusions] While introducing the dataset in the field of object detection, it also reveals the importance of object detection datasets, the challenges and characteristics in different scenarios, as well as the main challenges and future development trends in constructing object detection datasets.

Key words: object detection, dataset, pedestrian detection, face detection, computer vision

李林, 王家华, 周晨阳, 孔思曼, 孙践知. 目标检测数据集研究综述[J]. 数据与计算发展前沿, 2024, 6(2): 177-193.

LI Lin, WANG Jiahua, ZHOU Chenyang, KONG Siman, SUN Jianzhi. An Overview of Object Detection Datasets[J]. Frontiers of Data and Computing, 2024, 6(2): 177-193, https://cstr.cn/32002.14.jfdc.CN10-1649/TP.2024.02.016.

图/表 13

表1

图1

表2

图2

表3

人脸检测数据集"

数据集	发布年份	图像数量	实例数量	标注	清洗/ 检查	挑战性
AFLW^[27]	2011	21,997	25,993	边界框、关键点	手动/人工	面部姿势变化较大，光照和遮挡等因素影响较大
MALF^[28]	2014	5,250	11,931	边界框、性别、变形幅度、遮挡、佩戴眼镜、表情夸张、关键点	手动/人工	姿势变化、复杂背景
CelebA^[29]	2015	202,599	202,599	边界框、40个人脸属性、关键点	手动/人工	姿势变化、背景混乱
Wider Face^[30]	2015	32,203	393,703	边界框、遮挡、姿势、事件类别	手动/人工	主动选择的姿势、遮挡、表情等变化较大的人脸图像
IIIT-CFW^[31]	2016	8,927	—	边界框、类型、种族、性别、姿势、表情、外表、年龄、胡子、眼镜	手动/人工	卡通人脸识别、数据少、高度漫画化以及其它人脸检测挑战
MegaFace^[32]	2016	1.02M	—	无	自动/无	存在大量遮挡、未经标注
MS1M^[33]	2016	10M	—	无	-	规模空前庞大，未经清洗
UFDD^[34]	2017	6,425	10,897	边界框、天气、镜头障碍、模糊、光照变化、遮挡、关键点	手动/人工	关注雨、雪、霾等天气变化以及其它退化条件变化，包含大量干扰图像
Wildest Faces^[35]	2017	67,889	109,771	边界框、尺度、年龄、遮挡、姿势	手动/人工	暴力场景带来更狂野的表情，同时带来姿势巨大变化、遮挡和模糊的挑战
IJB-C^[36]	2018	31,334图像+11,779视频	89,642	边界框、性别、肤色、遮挡、粗姿势信息、关键点	手动/人工	多模态，光照变化剧烈、遮挡、低分辨率
VGG-Face2^[37]	2018	3.31M	—	边界框、姿势、年龄、关键点	半自动/人工	人脸多样性、标注质量低
MS1MV2^[38]	2019	5.8M	—	无	自动/无	标注质量不高、包含人脸检测中的广泛挑战
WebFace42M^[39]	2021	42M	—	边界框、姿势、年龄、种族、性别、帽子、眼镜和面具	自动/无	最大规模的公共人脸检测数据集，关注年龄、地区、背景等多层次的多样性

表3

图3

表4

表5

表6

图4

表7

图5

表8

文本检测数据集"

数据集	发布年份	图像数量	实例数量	标注	语言种类	文本方向	数据来源	挑战性
SVT^[68]	2009	350	725	单词边界框	英文	水平	谷歌街景图像	低分辨率、模糊、变形、复杂背景、光照不足
MSRA-TD500^[69]	2012	500	—	文本行边界框	中文、英文	多方向	相机拍摄	长文本图像、文本密度较高，背景复杂
IIIT 5k-word^[70]	2012	1,120	5,000	单词边界框、难度标注	英文	水平	谷歌图像搜索	包含场景文本和原生数字图像、存在光照变化、投影失真等挑战
ICDAR 2015^[71]	2015	1,500	11,886	单词边界框	英文	水平	可穿戴设备拍摄	大多数文本都没有焦点、图像存在严重失真或模糊
ICDAR 2017^[72]	2017	18,000	—	单词边界框	多语言	水平	手机拍摄及屏幕截图	自然场景，尺度、旋转、透视变换等方面更具挑战性
COCO Text^[73]	2017	63,686	173,589	单词边界框、位置、易读性、类别、文本脚本	多语言	多方向	MS COCO数据集	规模大，包含复杂日常场景的图像
Total-Text^[74]	2017	1,555	9,330	单词多边形边界框	英文	多方向	相机拍摄	关注曲线文本，近半曲线文本实例，存在遮挡、复杂背景挑战
MTWI^[75]	2018	20,000	—	文本行边界框、文本转录	多语言	多方向	淘宝网页	关注网络文本，包含复杂布局、水印、小文本、紧密堆叠文本和复杂形状文本等挑战
CTW^[76]	2019	32,285	1,018,402	文本行边界框、单词边界框、类别、遮挡、复杂背景、扭曲、凸起、艺术字、手写	中文	多方向	TT100K、腾讯街景	来自中国几十个城市的场景文本图像、场景多样性丰富，具有遮挡、背景复杂、文字扭曲、文字凸起、艺术字、手写字等挑战
ICDAR2019-MLT^[77]	2019	20,000	80,000	单词边界框	多语言	多方向	手机拍摄及互联网收集	自然场景、多语言

表8

参考文献 88

[1]	ZOU Z X, SHI Z W, GUO Y H, et al. Object Detection in 20 Years: A Survey[J]. Proceedings of the IEEE, 2019, 111(3):257-276. doi: 10.1109/JPROC.2023.3238524
[2]	PANDE B, PADAMWAR K, BHATTACHARYA S, et al. A Review of Image Annotation Tools for Object Detection[C]. 2022 International Conference on Applied Artificial Intelligence and Computing (ICAAIC), 2022: 976-982.
[3]	AGRAWAL A, JIASEN L, STANISLAW A, et al. VQA: Visual Question Answering[J]. International Journal of Computer Vision, 2017, 123(1): 4-31. doi: 10.1007/s11263-016-0966-6
[4]	GOYAL Y, TEJAS K, DOUGLAS S, et al. Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering[J]. International Journal of Computer Vision, 2016, 127(4): 398-414. doi: 10.1007/s11263-018-1116-0
[5]	EVERINGHAM M, GOOL L, WINN J, et al. The Pascal Visual Object Classes Challenge: A Retrospective[J]. International Journal of Computer Vision, 2014, 111(1): 98-136. doi: 10.1007/s11263-014-0733-5
[6]	LIN T Y, MAIRE M, HAYS J, et al. Microsoft COCO: Common Objects in Context[C]. European Conference on Computer Vision (ECCV), 2014: 740-755.
[7]	LI F F, FERGUS R, PERONA P. One-shot learning of object categories[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006, 28(4): 594-611. doi: 10.1109/TPAMI.2006.79
[8]	GRIFFIN G, HOLUB A, PERONA P. Caltech-256 Object Category Dataset[R]. California Institute of Technology, 2007.
[9]	RUSSELL B C, TORRALBA A, KEVIN P, et al. LabelMe: A Database and Web-Based Tool for Image Annotation[J]. International Journal of Computer Vision, 2007, 77(1-3): 157-173. doi: 10.1007/s11263-007-0090-8
[10]	DENG J, DONG W, SOCHER R, et al. ImageNet: A large-scale hierarchical image database[C]. 2009 IEEE Conference on Computer Vision and Pattern Recognition, 2009: 248-255.
[11]	XIAO J X, EHINGER K, HAYS J, et al. SUN Database: Exploring a Large Collection of Scene Categories[J]. International Journal of Computer Vision, 2014, 119(1): 3-22. doi: 10.1007/s11263-014-0748-y
[12]	KRISHNA R, ZHU Y, GROTH O, et al. Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations[J]. International Journal of Computer Vision, 2016, 123(1): 32-73. doi: 10.1007/s11263-016-0981-7
[13]	THOMEE B, SHAMMA D, FRIEDLAND G, et al. YFCC100M: The New Data in Multimedia Research[J]. Communications of the ACM, 2016, 59(2): 64-73.
[14]	KUZNETSOVA A, ROM H, ALLDRIN N, et al. The Open Images Dataset V4[J]. International Journal of Computer Vision, 2020, 128: 1956-1981. doi: 10.1007/s11263-020-01316-z
[15]	罗艳, 张重阳, 田永鸿, 等. 深度学习行人检测方法综述[J]. 中国图象图形学报, 2022, 27(7): 2094-2111.
[16]	LUO Y, ZHANG C, ZHAO M, et al. Where, What, Whether: Multi-modal Learning Meets Pedestrian Detection[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2020.
[17]	李晓艳, 符惠桐, 牛文涛, 等. 基于深度学习的多模态行人检测算法[J]. 西安交通大学学报, 2022, 56(10): 61-70.
[18]	ENZWEILER M, GAVRILA D. Monocular Pedestrian Detection: Survey and Experiments[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009, 31(12): 2179-2195. doi: 10.1109/TPAMI.2008.260 pmid: 19834140
[19]	DOLLÁR P, WOJEK C, SCHIELE B, et al. Pedestrian Detection: An Evaluation of the State of the Art[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(4): 743-761. doi: 10.1109/TPAMI.2011.155 pmid: 21808091
[20]	LI X F, FLOHR F, YUE Y, et al. A new benchmark for vision-based cyclist detection[C]. 2016 IEEE Intelligent Vehicles Symposium, 2016: 1028-1033.
[21]	CORDTS M, OMRAN M, RAMOS S, et al. The Cityscapes Dataset for Semantic Urban Scene Understanding[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition, 2016: 3213-3223.
[22]	BRAUN M, KREBS S, FLOHR F, et al. The EuroCity Persons Dataset: A Novel Benchmark for Object Detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2019, 41(8): 1844-1861. doi: 10.1109/TPAMI.34
[23]	CHOI Y, KIM N, HWANG S, et al. KAIST Multi-Spectral Day/Night Data Set for Autonomous and Assisted Driving[J]. IEEE Transactions on Intelligent Transportation Systems, 2018, 19(3): 934-948. doi: 10.1109/TITS.2018.2791533
[24]	SHAO S, ZHAO Z, LI B, et al. CrowdHuman: A Benchmark for Detecting Human in a Crowd[C]. arXiv preprint arXiv:1805.00123, 2018.
[25]	YU X H, GONG Y Q, JIANG N, et al. Scale Match for Tiny Person Detection[C]. 2020 IEEE Winter Conference on Applications of Computer Vision (WACV), 2020: 1246-1254.
[26]	李泽琛, 李恒超, 胡文帅, 等. 多尺度注意力学习的Faster R-CNN口罩人脸检测模型[J]. 西南交通大学学报, 2021, 56(5): 1002-1010.
[27]	KÖSTINGER M, WOHLHART P, ROTH P, et al. Annotated Facial Landmarks in the Wild: A large-scale, real-world database for facial landmark localization[C]. 2011 IEEE International Conference on Computer Vision Workshops (ICCV Workshops), 2011: 2144-2151.
[28]	YANG B, YAN J J, LEI Z, et al. Fine-grained evaluation on face detection in the wild[C]. 2015 11th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition, 2015: 1-7.
[29]	LIU Z, LUO P, WANG X, et al. Deep Learning Face Attributes in the Wild[C]. 2015 IEEE International Conference on Computer Vision (ICCV), 2015: 3730-3738.
[30]	YANG S, LUO P, LOY C C, et al. WIDER FACE: A Face Detection Benchmark[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 5525-5533.
[31]	MISHRA A, RAI S N, MISHRA A, et al. IIIT-CFW: A Benchmark Database of Cartoon Faces in the Wild[C]. European Conference on Computer Vision (ECCV), 2016: 35-47.
[32]	KEMELMACHER-SHLIZERMAN I, SEITZ S M, MILLER D, et al. The MegaFace Benchmark: 1 Million Faces for Recognition at Scale[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 4873-4882.
[33]	GUO Y D, ZHANG L, HU Y X, et al. MS-Celeb-1M: Challenge of Recognizing One Million Celebrities in the Real World[J]. electronic imaging, 2016, 2016(11): 1-6.
[34]	WANG L J, GAN Y K, LUO R X, et al. Unconstrained Face Detection via Anchor-based Region Proposal Network[C]. Proceedings of the 25th ACM International Conference on Multimedia, 2017: 1181-1189.
[35]	YUCEL M K, BILGE Y C, OGUZ O, et al. Wildest Faces: Face Detection and Recognition in Violent Settings[J]. arXiv preprint arXiv: 1805.07566, 2018.
[36]	MAZE B, ADAMS J C, DUNCAN J A, et al. IARPA Janus Benchmark-C: Face Dataset and Protocol[C]. 2018 International Conference on Biometrics (ICB), 2018: 158-165.
[37]	CAO Q, SHEN L, XIE W D, et al. VGGFace2: A Dataset for Recognising Faces across Pose and Age[C]. 2018 13th IEEE International Conference on Automatic Face & Gesture Recognition, 2018: 67-74.
[38]	DENG J, GUO J, ZAFEIRIOU J. ArcFace: Additive Angular Margin Loss for Deep Face Recognition[C]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019: 4685-4694.
[39]	ZHU Z, HUANG G, DENG J, et al. WebFace260M: A Benchmark Unveiling the Power of Million-Scale Deep Face Recognition[C]. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021: 10487-10497.
[40]	SERMANET P, LECUN Y. Traffic sign recognition with multi-scale Convolutional Networks[C]. The 2011 International Joint Conference on Neural Networks, 2011: 2809-2813.
[41]	胡均平, 王鸿树, 戴小标, 等. 改进YOLOv5的小目标交通标志实时检测算法[J]. 计算机工程与应用, 2023, 59(2): 185-193. doi: 10.3778/j.issn.1002-8331.2206-0503
[42]	HOUBEN S, STALLKAMP J, SALMEN J, et al. Detection of traffic signs in real-world images: The German traffic sign detection benchmark[C]. The 2013 International Joint Conference on Neural Networks (IJCNN), 2013: 1-8.
[43]	LARSSON F, FELSBERG M. Using Fourier Descriptors and Spatial Models for Traffic Sign Recognition[C]. Scandinavian Conference on Image Analysis, 2011: 238-249.
[44]	MØGELMOSE A, TRIVEDI M M, MOESLUND T B. Vision-Based Traffic Sign Detection and Analysis for Intelligent Driver Assistance Systems: Perspectives and Survey[J]. IEEE Transactions on Intelligent Transportation Systems, 2012, 13(4): 1484-1497. doi: 10.1109/TITS.2012.2209421
[45]	TIMOFTE R, ZIMMERMANN K, GOOL L. Multi-view traffic sign detection, recognition, and 3D localization[J]. Machine Vision and Applications, 2009, 25(3): 633-647. doi: 10.1007/s00138-011-0391-3
[46]	ZHU Z, LIANG D, ZHANG S H, et al. Traffic-Sign Detection and Classification in the Wild[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 2110-2118.
[47]	ZHANG J M, HUANG M T, JIN X K, et al. A Real-Time Chinese Traffic Sign Detection Algorithm Based on Modified YOLOv2[J]. Algorithms, 2017, 10(4): 127. doi: 10.3390/a10040127
[48]	TEMEL D, ALSHAWI T A, CHEN M, et al. Challenging Environments for Traffic Sign Detection: Reliability Assessment under Inclement Conditions[J]. arXiv preprint arXiv: 1902.06857, 2019.
[49]	钱伍, 王国中, 李国平. 改进YOLOv5的交通灯实时检测鲁棒算法[J]. 计算机科学与探索, 2022, 16(1): 231-241. doi: 10.3778/j.issn.1673-9418.2105033
[50]	DE CHARETTE R, NASHASHIBI F. Real time visual traffic lights recognition based on Spot Light Detection and adaptive traffic lights templates[C]. 2009 IEEE Intelligent Vehicles Symposium, 2009: 358-363.
[51]	JENSEN MB, PHILIPSEN MP, MØGELMOSE A, et al. Vision for Looking at Traffic Lights: Issues, Survey, and Perspectives[J]. IEEE Transactions on Intelligent Transportation Systems, 2016, 17(7): 1800 -1815. doi: 10.1109/TITS.2015.2509509
[52]	FREGIN A, MÜLLER J, KREBEL U, et al. The DriveU Traffic Light Dataset: Introduction and Comparison with Existing Datasets[C]. 2018 IEEE International Conference on Robotics and Automation (ICRA), 2018:3376-3383.
[53]	BEHRENDT K, NOVAK L, BOTROS R. A deep learning approach to traffic lights: Detection, tracking, and classification[C]. 2017 IEEE International Conference on Robotics and Automation (ICRA), 2017: 1370-1377.
[54]	YU F, CHEN H, WANG X, et al. BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learning[C]. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020:2633-2642.
[55]	PINGGERA P, RAMOS S, GEHRIG S K, et al. Lost and Found: detecting small road hazards for self-driving vehicles[C]. 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2016: 1099-1106.
[56]	PAN X G, SHI J P, LUO P, et al. Spatial As Deep: Spatial CNN for Traffic Scene Understanding[C]. AAAI Conference on Artificial Intelligence, 2018:7276: 7283.
[57]	YOO S, LEE H, MYEONG H, et al. End-to-End Lane Marker Detection via Row-wise Classification[C]. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2020: 4335-4343.
[58]	BEHRENDT K AND SOUSSAN R. Unsupervised Labeled Lane Markers Using Maps[C]. 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW), 2019: 832-839.
[59]	赵文清, 孔子旭, 周震东, 等. 增强小目标特征的航空遥感目标检测[J]. 中国图象图形学报, 2021, 26(3): 644-653.
[60]	LIU K, MÁTTYUS G. Fast Multiclass Vehicle Detection on Aerial Images[J]. IEEE Geoscience and Remote Sensing Letters, 2015, 12(9): 1938-1942. doi: 10.1109/LGRS.2015.2439517
[61]	RAZAKARIVONY S, JURIE F. Vehicle detection in aerial imagery : A small target detection benchmark[J]. Journal of Visual Communication and Image Representation, 2016, 34: 187-203. doi: 10.1016/j.jvcir.2015.11.002
[62]	MUNDHENK T N, KONJEVOD G, SAKLA W A, et al. A Large Contextual Dataset for Classification, Detection and Counting of Cars with Deep Learning[C]. European Conference on Computer Vision (ECCV), 2016: 785-800.
[63]	XIA G S, BAI X, DING J, ET AL. DOTA: A Large-Scale Dataset for Object Detection in Aerial Images[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 3974-3983.
[64]	LAM D, KUZMA R, MCGEE K, et al. xView: Objects in Context in Overhead Imagery[J]. arXiv preprint arXiv: 1802.07856, 2018.
[65]	ZHU P F, WEN L Y, BIAN X, et al. Vision Meets Drones: A Challenge[J]. arXiv preprint arXiv:1804.07437, 2018.
[66]	LI K, WAN G, CHENG G, et al. Object Detection in Optical Remote Sensing Images: A Survey and A New Benchmark[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2019, 159: 296-307. doi: 10.1016/j.isprsjprs.2019.11.023
[67]	DING J, XUE N, XIA G, et al. Object Detection in Aerial Images: A Large-Scale Benchmark and Challenges[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(11): 7778-7796. doi: 10.1109/TPAMI.2021.3117983
[68]	WANG K, BELONGIE S J. Word Spotting in the Wild[C]. European Conference on Computer Vision (ECCV), 2010: 591-604.
[69]	YAO C, BAI X, LIU W Y, et al. Detecting texts of arbitrary orientations in natural images[C]. 2012 IEEE Conference on Computer Vision and Pattern Recognition, 2012: 1083-1090.
[70]	MISHRA A, KARTEEK A, JAWAHAR C. Scene Text Recognition using Higher Order Language Priors[C]. British Machine Vision Conference (BMVC), 2012.
[71]	KARATZAS D, BIGORDA L G, NICOLAOU A, et al. ICDAR 2015 competition on Robust Reading[C]. 2015 13th International Conference on Document Analysis and Recognition (ICDAR), 2015: 1156-1160.
[72]	NAYEF N, YIN F, BIZID I, et al. ICDAR2017 Robust Reading Challenge on Multi-Lingual Scene Text Detection and Script Identification - RRC-MLT[C]. 2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR), 2017: 1454-1459.
[73]	VEIT A, MATERA T, NEUMANN L, et al. COCO-Text: Dataset and Benchmark for Text Detection and Recognition in Natural Images[J]. arXiv preprint arXiv:1601.07140, 2016.
[74]	CHNG C AND CHAN C. Total-Text: A Comprehensive Dataset for Scene Text Detection and Recognition[C]. 2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR), 2017: 935-942.
[75]	HE M, LIU Y, YANG Z, et al. ICPR2018 Contest on Robust Reading for Multi-Type Web Images[C]. 2018 24th International Conference on Pattern Recognition (ICPR), 2018: 7-12.
[76]	YUAN T, ZHU Z, XU K, et al. A Large Chinese Text Dataset in the Wild[J]. Journal of Computer Science and Technology, 2019, 34(3): 509-521. doi: 10.1007/s11390-019-1923-y
[77]	NAYEF N, PATEL Y, BUSTA M, et al. ICDAR2019 Robust Reading Challenge on Multi-lingual Scene Text Detection and Recognition — RRC-MLT-2019[C]. 2019 International Conference on Document Analysis and Recognition (ICDAR), 2019: 1582-1587.
[78]	CUBUK E D, ZOPH B, MANÉ D, et al. AutoAugment: Learning Augmentation Strategies From Data[C]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019: 113-123.
[79]	SHORTEN C, KHOSHGOFTAAR M. A survey on Image Data Augmentation for Deep Learning[J]. Journal of Big Data, 2019, 6(1): 1-48. doi: 10.1186/s40537-018-0162-3
[80]	ZHANG H Y, CISSÉ M, DAUPHIN Y, et al. mixup: Beyond Empirical Risk Minimization[J]. arXiv preprint arXiv:1710.09412, 2017.
[81]	BERTHELOT D, CARLINI N, GOODFELLOW I J, et al. MixMatch: A Holistic Approach to Semi-Supervised Learning[J]. arXiv preprint arXiv:1905.02249, 2019.
[82]	LAINE S, AILA T. Temporal Ensembling for Semi-Supervised Learning[J]. arXiv preprint arXiv:1610.02242, 2016.
[83]	DOERSCH C, GUPTA A K, EFROS A A. Unsupervised Visual Representation Learning by Context Prediction[C]. 2015 IEEE International Conference on Computer Vision (ICCV), 2015: 1422-1430.
[84]	QI C, LIU W, WU C X, et al. Frustum PointNets for 3D Object Detection from RGB-D Data[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2017: 918-927.
[85]	SHI S S, WANG X G, LI H S. PointRCNN: 3D Object Proposal Generation and Detection From Point Cloud[C]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2018:770-779.
[86]	ZHOU Y, TUZEL O. VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 4490-4499.
[87]	FIEDLER N, BESTMANN M, HENDRICH N. ImageTagger: An Open Source Online Platform for Collaborative Image Labeling[M]. RoboCup 2018:Robot World Cup XXII. Lecture Notes in Computer Science, 2019, 11374: 162-169.
[88]	LAHTINEN T, TURTIAINEN H, COSTIN A I. Brima: Low-Overhead Browser-Only Image Annotation Tool (Preprint)[C]. 2021 IEEE International Conference on Image Processing (ICIP), 2021: 2633-2637.

数据集	发布年份	图像数量	类别数量	特点
VOC2007/2012	2007/2012	9,963/11,540	20/20	大小、方向、姿势、光照、位置、遮挡等方面具有较大变化
MS COCO	2014	328,000	91	小目标较多、单幅图片目标多，大多数类别物体对应较多的实例
Caltech 101/256	2006/2007	9,146/30,607	101/256	大多数图像仅包含一个物体，图像变化较少
LabelMe	2008	187,240	1,000+	提供了在线注释工具，更具开放性
ImageNet	2009	14M	21,841	数据集规模大、图像类别较多，具有很大挑战性
SUN	2010	131,072	908	具有丰富的场景分类，对图像上的场景和对象提供了完整注释
YFCC100M	2014	99.2M图像+800K视频	—	包含图像和视频的多模态数据集，数据集规模十分庞大
VQAv1/v2	2015/2017	254,721/286,046	—	在MS COCO的基础上形成的以视觉问答为主的图像数据集
Visual Genome	2016	108,249	76,340	多模态数据集，融合了区域描述、关系、问题答案对等多种模态信息
Open ImagesV4	2018	1.9M	600	手工注释，图像多显示为多目标的复杂场景

数据集	发布年份	图像数量	实例数量	分辨率	标注	城市数量	行人密度	数据来源	挑战性
Daimler-DB^[18]	2008	52910	60,407	640×480	边界框	1	低	车载摄像机	遮挡、尺度变化较大
Caltech Pedestrian^[19]	2009	250000	350,000	640×480	边界框、时间对应关系	1	低	车载摄像机	遮挡、多尺度
TDC^[20]	2016	14674	32,361	2048×1024	边界框、行人属性	1	低	车载摄像机	关注骑行者、遮挡实例非常多、存在极端遮挡情况
CityPersons^[21]	2017	5000	35,016	2048×1024	边界框、行人属性	27	中等	监控视频、互联网采集	季节、场景多样性，遮挡、运动模糊、行人密集
EuroCity^[22]	2018	47337	238,285	1920×1024	边界框、行人属性	31	中等	车载摄像机	季节、城市场景多样，光照变化剧烈、运动模糊、行人密集
KAIST^[23]	2018	95328对	103,128	不定	边界框	1	低	监控视频、车载摄像机	多模态、场景丰富、光照变化大、形态多样性
CrowdHuman^[24]	2018	24000	470,000	不定	身体和头部边界框、行人属性	40+	高	互联网采集	丰富的场景、遮挡、极大的人群密度、形态多样性
Tiny Person^[25]	2019	1610	72,651	不定	边界框	—	高	互联网采集	远距离、大量背景干扰、关注微小目标、极大的人群密度

数据集	发布年份	图像数量	实例数量	类别数量	分辨率	标注	场景	数据来源	挑战性
GTSDB^[42]	2011	900	1,206	4	1360×800	类型、边界框	白天、黄昏、城市、农村、高速公路	德国高速公路	光照变化、遮挡、模糊、阴影、雾霾、阴天、雨天
STS^[43]	2011	4000	3,488	7	1280×960	类型、可见状态、道路状态、边界框	白天、城市、高速公路	瑞典道路	小目标、光照、遮挡、阴影、模糊、阴天、雨天
LISA^[44]	2012	6,610图像, 17视频	7,855	49	640×480~1024×522	类型、遮挡、道路状态、边界框	白天、夜晚、城市、农村、高速公路	美国加州道路	多模态、小目标、背景复杂、光照、遮挡、阴影、模糊、镜头污垢、阴天
BelgiumTSC^[45]	2013	7,356	11,219	62	640×480	类型、边界框、3D位置	白天、高速公路	比利时公路	多视角、光照、遮挡、变形、小目标
TT100K^[46]	2016	100,000	30,000	128	2048×2048	类型、边界框、边界顶点	白天、夜晚、市中心、郊区	腾讯地图记录的中国5个城市街景数据	光照、天气、尺度的巨大变化、遮挡、小目标
CCTSDB^[47]	2017	10,000	13,361	3	1000×350	类型、边界框	白天、城市、高速公路	中国道路	小目标、背景复杂、光照、遮挡、阴影、雨天、阴天、镜头污垢、雾霾、模糊
CURE-TSD^[48]	2019	1,719,900图像,5,733视频	2,206,106	14	1628×1236	类型、边界框、挑战类型和等级	多种场景	比利时真实数据和合成虚拟数据	多模态、阴天、雨天、雪天、阴影、雾霾、光照、模糊、镜头污垢、遮挡

数据集	发布年份	图像数量	实例数量	类别	分辨率	标注	数据来源	挑战性
LaRA^[50]	2013	11,179	9,168	4	640×480	边界框、类别	法国巴黎，车载相机拍摄	场景和天气条件具有多样性、存在遮挡和模糊情况
LISA^[51]	2016	43,007图像+18视频	113,888	7	1280×960	边界框、类别、状态	美国加州，车载相机拍摄	不同的光照条件、天气条件多样、场景丰富
DTLD^[52]	2016	3,366图像+188视频	230,000	344	2048×1024	边界框、类别、方向	德国11个城市，车载相机拍摄	规模大、交通灯分类多、小目标非常多、天气条件多样
BSTLD^[53]	2017	5,093/8,334	10,756/13,493	15/4	1280×720	边界框、类别、状态	美国加州，车载相机拍摄	阴影、光照、小目标
BDD100K^[54]	2020	100K视频，100K图像	265,906	4	1280×720	边界框、类别、天气、场景、时间	众包方式获得，由全球各地司机上传	规模大、光照变化、具有雨雪等不同的恶劣天气条件、场景多样

数据集	发布时间	图片数量	分辨率	标注	应用场景	数据来源	特点
Lost and Found^[55]	2016	21,000图像+112视频	1920×1080	道路障碍物类别、ID	障碍物检测	街道、车载相机	包含多种街道场景和障碍物类型，目标存在大小、距离、颜色和材料等变化，具有不规则的道路轮廓、较远的目标距离、不同的道路表面外观、强烈的光照变化等挑战
CULane^[56]	2017	133,235	1640×590	车道标注、上下文注释	车道检测	北京，车载相机	包含人流量大、光线暗、无车线道路、阴影、眩光、弯道、岔道、车道遮挡严重、道路状况复杂等挑战
TuSimple lane^[57]	2018	6,408	1280×720	车道标注	车道检测	北京，车载相机	车道线以点来标注，存在眩光、遮挡等挑战
LLAMAS^[58]	2019	100,042	1280×717	2D、3D虚线车道标记的像素级注释	车道检测	高速公路，车载相机	基于高精度地图实现的无监督车道标记数据集，数据集进行了自动标注、人工检查