自适应聚焦损失的图像目标检测算法

doi:10.3778/j.issn.1002-8331.2104-0321

摘要/Abstract

摘要：

现代目标检测算法仍然存在由现有目标检测架构引起的正负样本不平衡和训练数据引起的难易样本不平衡。现有方法一般采用基于类别频率的重采样或基于类别预测概率的重新加权，虽然减轻了类别的不平衡问题，但是引入了新的超参数，为每个训练任务需要进行大量的手动调整超参数。为此在现有Focal Loss损失函数基础上提出了一个新的损失函数自适应聚焦损失（Adaptive Focal Loss），使模型聚焦于对训练过程贡献更大的困难样本，并且可自适应地调整超参数。根据训练过程中每批图像标签中的正负样本数量计算出自适应的加权因子来实现对正负样本的动态平衡。根据训练过程中不同阶段各类真实标签的期望概率计算出自适应的调制因子来自适应地平衡难易样本。为验证方法的有效性，在PASCAL VOC2007测试数据集中平均精度均值达到80.75%，相比较于原算法提高了3.45个百分点。在PASCAL VOC2012测试数据集中平均精度均值达到77.17%，相比较于原算法提高了1.87个百分点。实验结果表明，把Adaptive Focal Loss作为网络的损失函数，相比于原始的Focal Loss损失函数检测精度有所提升，并具有较大的实用价值。

关键词: 目标检测, 样本不平衡, 自适应聚焦损失

Abstract:

Modern target detection algorithms still have the imbalance of positive and negative samples caused by the existing target detection architecture and the imbalance of hard and easy samples caused by the training data. The existing methods generally use resampling based on class frequency or reweighting based on class prediction probability. Although the imbalance of classes is alleviated, new super parameters are introduced, which requires a lot of manual adjustment for each training task. For this reason, a new loss function Adaptive Focal Loss is proposed on the basis of the existing Focal Loss, which makes the model focus on the more difficult samples and adjust the super parameters adaptively. Firstly, according to the number of positive and negative samples in each batch of image tags in the training process, the self-adaptive weighting factor is calculated to achieve the dynamic balance of positive and negative samples. Secondly, according to the expected probabilities of all kinds of ground-truth label in different stages of the training process, the adaptive modulation factor is calculated, and the adaptive balanced samples are obtained. In order to verify the effectiveness of the method, the mAP of PASCAL VOC2007 test data set reaches 80.75%, which is 3.45 percentage points higher than the original algorithm. In PASCAL VOC2012 test data set, the mAP is 77.17%, which is 1.87 percentage points higher than the original algorithm. The experimental results show that, compared with the original Focal Loss, the detection accuracy of Adaptive Focal Loss is improved, and it has great practical value.

Key words: object detection, sample imbalance, adaptive focal loss

肖振久，孔祥旭，宗佳旭，杨玥莹. 自适应聚焦损失的图像目标检测算法[J]. 计算机工程与应用, 2021, 57(23): 185-192.

XIAO Zhenjiu, KONG Xiangxu, ZONG Jiaxu, YANG Yueying. Image Object Detection Algorithm Based on Adaptive Focal Loss[J]. Computer Engineering and Applications, 2021, 57(23): 185-192.

参考文献

[1] HE K，GKIOXARI G，DOLLáR P，et al.Mask R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision，2017：2961-2969.
[2] LIU W，ANGUELOV D，ERHAN D，et al.SSD：single shot multibox detector[C]//European Conference on Computer Vision.Cham：Springer，2016：21-37.
[3] EVERINGHAM M，GOOL L V，WILLIAMS C，et al.The pascal visual object classes（VOC） challenge[J].International Journal of Computer Vision，2010，88（2）：303-338.
[4] LIN T Y，MAIRE M，BELONGIE S，et al.Microsoft coco：common objects in context[C]//European Conference on Computer Vision.Springer：Cham，2014：740-755.
[5] REN S，HE K，GIRSHICK R，et al.Faster R-CNN：towards real-time object detection with region proposal networks[J].arXiv：1506.01497，2015.
[6] UIJLINGS J R R，VAN DE SANDE K E A，GEVERS T，et al.Selective search for object recognition[J].International Journal of Computer Vision，2013，104（2）：154-171.
[7] ZITNICK C L，DOLLáR P.Edge boxes：locating object proposals from edges[C]//European Conference on Computer Vision.Cham：Springer，2014：391-405.
[8] PINHEIRO P O，COLLOBERT R，DOLLáR P.Learning to segment object candidates[J].arXiv：1506.06204，2015.
[9] GIRSHICK R，DONAHUE J，DARRELL T，et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2014：580-587.
[10] SHRIVASTAVA A，GUPTA A，GIRSHICK R.Training region-based object detectors with online hard example mining[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2016：761-769.
[11] LIN T Y，GOYAL P，GIRSHICK R，et al.Focal loss for dense object detection[C]//Proceedings of the IEEE International Conference on Computer Vision，2017：2980-2988.
[12] LECUN Y，BOTTOU L，BENGIO Y，et al.Gradient-based learning applied to document recognition[J].Proceedings of the IEEE，1998，86（11）：2278-2324.
[13] LONG J，SHELHAMER E，DARRELL T.Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2015：3431-3440.
[14] KRIZHEVSKY A，SUTSKEVER I，HINTON G E.Imagenet classification with deep convolutional neural networks[C]//Advances in Neural Information Processing Systems，2012：1097-1105.
[15] GIRSHICK R.Fast R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision，2015：1440-1448.
[16] DAI J，LI Y，HE K，et al.R-fcn：object detection via region-based fully convolutional networks[J].arXiv：1605.06409，2016.
[17] SERMANET P，EIGEN D，ZHANG X，et al.Overfeat：integrated recognition，localization and detection using convolutional networks[J].arXiv：1312.6229，2013.
[18] FU C Y，LIU W，RANGA A，et al.Dssd：deconvolutional single shot detector[J].arXiv：1701.06659，2017.
[19] REDMON J，DIVVALA S，GIRSHICK R，et al.You only look once：unified，real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2016：779-788.
[20] REDMON J，FARHADI A.YOLO9000：better，faster，stronger[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2017：7263-7271.
[21] REDMON J，FARHADI A.Yolov3：an incremental improvement[J].arXiv：1804.02767，2018.
[22] OKSUZ K，CAM B C，KALKAN S，et al.Imbalance problems in object detection：a review[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2021，43（10）：3388-3415.
[23] BULO S R，NEUHOLD G，KONTSCHIEDER P.Loss max-pooling for semantic image segmentation[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition，2017：7082-709.
[24] HASTIE T，TIBSHIRANI R，TIBSHIRANI R，et al.The elements of statistical learning[M].Berlin：Springer，2001.
[25] LI B，LIU Y，WANG X.Gradient harmonized single-stage detector[C]//Proceedings of the AAAI Conference on Artificial Intelligence，2019：8577-8584.
[26] Wang X，Peng Y，Lu L，et al.Hospital-scale chest x-ray database and benchmarks on weakly-supervised classification and localization of common thorax diseases[C]//IEEE CVPR，2017：3462-3471.
[27] JEONG J，PARK H，KWAK N.Enhancement of SSD by concatenating feature maps for object detection[J].arXiv：1705.09587，2017.
[28] CHOI J，LEE K，JEONG J，et al.Two-layer residual feature fusion for object detection[C]//ICPRAM，2019：352-359.
[29] ROMERA E，BERGASA L M，ALVAREZ J M，et al.Train here，deploy there：robust segmentation in unseen domains[C]//2018 IEEE Intelligent Vehicles Symposium（IV），2018：1828-1833.
[30] SIMONYAN K，ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[J].arXiv：1409.1556，2014.
[31] HE K，ZHANG X，REN S，et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2015，37（9）：1904-1916.
[32] 宋艳艳，谭励，马子豪，等.改进 YOLOV3 算法的视频目标检测[J].计算机科学与探索，2021，15（1）：163-172.
SONG Y Y，TAN L，MA Z H，et al.Video target detection based on improved YOLOV3 algorithm[J].Journal of Frontiers of Computer Science and Technology，2021，15（1）：163-172.
[33] 金静，党建武，王阳萍，等.自适应低秩稀疏分解在运动目标检测中的应用[J].计算机科学与探索，2016，10（12）：1744-1751.
JING J，DANG J W，WANG Y P，et al.Application of adaptive low-rank and sparse decomposition in moving objections detection[J].Journal of Frontiers of Computer Science and Technology，2016，10（12）：1744-1751
[34] 梁延禹，李金宝.多尺度非局部注意力网络的小目标检测算法[J].计算机科学与探索，2020，14（10）：1744-1753.
LIANG Y Y，LI J B.Small objects detection method based on multi-scale non-local attention network[J].Journal of Frontiers of Computer Science and Technology，2020，14（10）：1744-1753.
[35] 宋欢欢，惠飞，景首才，等.改进的RetinaNet模型的车辆目标检测[J].计算机工程与应用，2019，55（13）：225-230.
SONG H H，HUI F，JING S C，et al.Improved Retina-Net model for vehicle target detection[J].Computer Engineering and Applications，2019，55（13）：225-230.
[36] 陈灏然，彭力，李文涛，等.加权网络下的小目标检测算法[J/OL].计算机科学与探索（2021-03-29）[2021-04-10].http://kns.cnki.net/kcms/detail/11.5602.TP.20210326.1129.
005.html.
CHEN H R，PENG L，LI W T.Small object detection algorithm based on weighted network[J/OL].Journal of Frontiers of Computer Science and Technology（2021-03-29）[2021-04-10].http：//kns.cnki.net/kcms/detail/11.5602.
TP.20210326.1129.005.html.
[37] PINHEIRO P O，LIN T Y，COLLOBERT R，et al.Learning to refine object segments[C]//European Conference on Computer Vision.Cham：Springer，2016：75-91.
[38] LIN T Y，DOLLáR P，GIRSHICK R，et al.Feature pyramid networks for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2017：2117-2125.
[39] LI Z，PENG C，YU G，et al.Light-head R-CNN：in defense of two-stage object detector[J].arXiv：1711.07264，2017.