基于分类不确定性的伪标签目标检测算法

doi:10.19678/j.issn.1000-3428.0063678

摘要/Abstract

摘要： 伪标签目标检测算法利用大量未标注数据生成伪标签数据来增加训练数据规模，从而提高目标检测模型的性能。针对伪标签数据中存在大量错误标注数据且伪标签目标检测模型性能难以提升的问题，提出基于SoftTeacher-CUC的伪标签目标检测算法。SoftTeacher-CUC算法在SoftTeacher伪标签目标检测算法的基础上，利用分类不确定性方法计算模型生成的伪标签分类结果的不确定性来判断伪标签是否可靠，不确定性越低说明伪标签的分类结果越可靠。在此基础上，将计算得到的不确定性作为权重加入伪标签数据的分类损失函数中，进一步减少高不确定性伪标签为模型带来的负面影响。根据Teacher模型中不同模块的作用，采用不同权重的指数滑动平均方法更新Teacher模型，降低Teacher模型和Student模型参数之间的相似性，使一致性正则化方法发挥效用。实验结果表明，在标注数据分别占训练集1%、5%和10%的情况下，与SoftTeacher算法相比，SoftTeacher-CUC算法的平均精度均值分别提高了1.4、1.2和1.7个百分点，在标注数据较少的情况下，该算法具有更好的检测效果。

关键词: 目标检测, 伪标签, 分类不确定性, 指数滑动平均, 分类损失函数, 一致性正则化

Abstract: The pseudo-label object detection algorithm aims to increase the training data size by using a large amount of unlabeled data to generate pseudo-label data in order to improve the performance of the object detection model.To mitigate the problem of a large amount of incorrectly labeled data in the pseudo-label data and that the performance of the pseudo-label object detection model is difficult to improve, this paper proposes a pseudo-label object detection algorithm of SoftTeacher-CUC.The SoftTeacher-CUC algorithm is based on the SoftTeacher pseudo-label object detection algorithm.First, the classification uncertainty calculation method calculates the uncertainty of the pseudo-label classification results generated by the model to determine whether the pseudo-label is reliable.The lower the uncertainty, the more reliable the pseudo-label classification results.Then, the calculated uncertainty is added as a weight to the classification loss of the pseudo-label data to further reduce the negative impact of high-uncertainty pseudo-labels on the model.Finally, according to the role of the different modules of the Teacher model, an Exponential Moving Average(EMA) method with different weights is used to update the Teacher model.It reduces the similarity between the parameters of the Teacher model and those of the Student model to enable the consistency regularization method to work.The experimental results show that the labeled data accounts for 1%, 5%, and 10% of the training set.Compared with the SoftTeacher algorithm, the mAP of the SoftTeacher-CUC algorithm is improved by 1.4, 1.2, and 1.7 percentage points, respectively.Consequently, SoftTeacher-CUC algorithm has a better detection effect when there is less labeled data.

Key words: object detection, pseudo-label, classification uncertainty, Exponential Moving Average(EMA), classification loss function, consistency regularization

中图分类号:

TP391

雷洁, 饶文碧, 杨焱超, 熊盛武. 基于分类不确定性的伪标签目标检测算法[J]. 计算机工程, 2023, 49(1): 49-56.

LEI Jie, RAO Wenbi, YANG Yanchao, XIONG Shengwu. Pseudo-Label Object Detection Algorithm Based on Classification Uncertainty[J]. Computer Engineering, 2023, 49(1): 49-56.

http://www.ecice06.com/CN/Y2023/V49/I1/49

图/表 7

20230701175036

20230701175040

20230701175043

20230701175047

20230701175050

20230701175053

20230701175057

参考文献

[1] REN S Q, HE K M, GIRSHICK R, et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149.
[2] LIN T Y, GOYAL P, GIRSHICK R, et al.Focal loss for dense object detection[C]//Proceedings of IEEE International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2017:2999-3007.
[3] LIU W.SSD:single shot multibox detector[C]//Proceedings of European Conference on Computer Vision.Berlin, German:Springer, 2016:21-37.
[4] REDMON J, FARHADI A.YOLO9000:better, faster, stronger[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2017:6517-6525.
[5] RIZVE M N, DUARTE K, RAWAT Y S, et al.In defense of pseudo-labeling:an uncertainty-aware pseudo-label selection framework for semi-supervised learning[EB/OL].[2021-11-20].https://arxiv.org/abs/2101.06329.
[6] LIX, SUN Q, LIU Y, et al.Learning to self-train for semi-supervised few-shot classification[C]//Proceedings of Annual Conference on Neural Information Processing Systems.Washington D.C., USA:IEEE Press, 2019:10276-10286.
[7] XIE Q Z, LUONG M T, HOVY E, et al.Self-training with noisy student improves ImageNet classification[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2020:10684-10695.
[8] 尹玉, 詹永照, 姜震.伪标签置信选择的半监督集成学习视频语义检测[J].计算机应用, 2019, 39(8):2204-2209. YIN Y, ZHAN Y Z, JIANG Z.Semi-supervised ensemble learning for video semantic detection based on pseudo-label confidence selection[J].Journal of Computer Applications, 2019, 39(8):2204-2209.(in Chinese)
[9] 史蕴豪, 许华, 刘英辉.一种基于伪标签半监督学习的小样本调制识别算法[J].西北工业大学学报, 2020, 38(5):1074-1083. SHI Y H, XU H, LIU Y H.A few-shot modulation recognition method based on pseudo-label semi-supervised learning[J].Journal of Northwestern Polytechnical University, 2020, 38(5):1074-1083.(in Chinese)
[10] SOHN K, ZHANG Z Z, LI C L, et al.A simple semi-supervised learning framework for object detection[EB/OL].[2021-11-20].https://arxiv.org/abs/2005.04757.
[11] ZHOU Q, YU C H, WANG Z B, et al.Instant-teaching:an end-to-end semi-supervised object detection framework[C]//Proceedings of 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2021:4079-4088.
[12] LIU Y C, MA C Y, HE Z J, et al.Unbiased teacher for semi-supervised object detection[EB/OL].[2021-11-20].https://arxiv.org/abs/2102.09480.
[13] PHAM H, DAI Z H, XIE Q Z, et al.Meta pseudo labels[C]//Proceedings of 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2021:11552-11563.
[14] WANG Y K, XU C M, LIU C, et al.Instance credibility inference for few-shot learning[C]//Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2020:12833-12842.
[15] ISCEN A, TOLIAS G, AVRITHIS Y, et al.Label propagation for deep semi-supervised learning[C]//Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2019:5065-5074.
[16] SOHN K, BERTHELOT D, LI C L, et al.FixMatch:simplifying semi-supervised learning with consistency and confidence[EB/OL].[2021-11-20].https://arxiv.org/abs/2001.07685.
[17] BERTHELOT D, CARLINI N, GOODFELLOW I, et al.MixMatch:a holistic approach to semi-supervised learning[EB/OL].[2021-11-20].https://arxiv.org/abs/1905.02249.
[18] YANG X L, SONG Z X, KING I, et al.A survey on deep semi-supervised learning[J].IEEE Transactions on Knowledge and Data Engineering, 2022, 36(7):1-20.
[19] XU M D, ZHANG Z, HU H, et al.End-to-end semi-supervised object detection with soft teacher[C]//Proceedings of 2021 IEEE/CVF International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2021:3040-3049.
[20] FOONG A Y K, BURT D R, LI Y Z, et al.On the expressiveness of approximate inference in Bayesian neural networks[EB/OL].[2021-11-20].https://arxiv.org/abs/1909.00719.
[21] WANG H, YEUNG D Y.Towards Bayesian deep learning:a framework and some existing methods[J].IEEE Transactions on Knowledge and Data Engineering, 2016, 28(12):3395-3408.
[22] GAL Y, GHAHRAMANI Z.Dropout as a Bayesian approximation:representing model uncertainty in deep learning[EB/OL].[2021-11-20].https://arxiv.org/abs/1506.02142.
[23] BRACH K, SICK B, DÜRR O.Single shot MC dropout approximation[EB/OL].[2021-11-20].https://arxiv.org/abs/2007.03293.
[24] BERTONI L, KREISS S, ALAHI A.MonoLoco:monocular 3D pedestrian localization and uncertainty estimation[C]//Proceedings of 2019 IEEE/CVF International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2019:6860-6870.
[25] ABDAR M.A review of uncertainty quantification in deep learning:techniques, applications and challenges[J].Information Fusion, 2021, 76:243-297.
[26] CAI Z W, RAVICHANDRAN A, MAJI S, et al.Exponential moving average normalization for self-supervised and semi-supervised learning[C]//Proceedings of 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2021:194-203.

选择文件类型/文献管理软件名称

选择包含的内容