Improved Algorithm for Fully Convolutional Object Detection

doi:10.3778/j.issn.1002-8331.2111-0441

Abstract

Abstract: The anchor-free fully convolutional one-stage objectdetection（FCOS） avoids the problem of sample imbalance because it does not generate a large number of anchors. However, FCOS may be applicable to specific scenario. To improve feature fusion and accuracy of object detection, an improved algorithm for fully convolutional object detection named ConFCOS is proposed, which designs a strengthened feature pyramid network（SFPN） with the attention modules of global context information and dilated convolutional modules to reduce information attenuation in feature fusion. Moreover, a cascade detection head is constructed to detect the objects, which uses the cascade detection to refine the bounding box regression and improve the confidence of classification and regression. In addition, the loss function of ConFCOS is optimized to accelerate model training and improve performance of detector. Experimental results on COCO datasets show that the mAP of ConFCOS is about 1.6?percentage points higher than FCOS.

Key words: ConFCOS, strengthened feature pyramid network（SFPN）, cascade detection, object detection

摘要： 基于无锚点的单阶段全卷积目标检测算法（FCOS）无需生成大量的锚点避免了样本不平衡问题，但FCOS可能更适应于某一特定场景。为了增强特征融合，并提高目标检测的准确性，提出了全卷积目标检测算法FCOS的改进算法ConFCOS。该算法设计了一个增强的特征金字塔网络，引入带有全局上下文信息的注意力模块和空洞卷积模块，以减少特征融合过程中的信息衰减。另外，构建了一个级联检测头来检测对象，对检测的边界框进行细化来提高分类和回归的置信度。此外，针对提出的ConFCOS的损失函数进行了优化以提高目标检测的准确率。在COCO数据集上进行的实验表明，ConFCOS的准确度比FCOS提高了1.6个百分点。

关键词: ConFCOS, 增强的特征金字塔网络, 级联检测, 目标检测

LIAO Yongwei, ZHANG Guipeng, YANG Zhenguo, LIU Wenyin. Improved Algorithm for Fully Convolutional Object Detection[J]. Computer Engineering and Applications, 2022, 58(17): 158-164.

廖永为, 张桂鹏, 杨振国, 刘文印. 全卷积目标检测的改进算法[J]. 计算机工程与应用, 2022, 58(17): 158-164.

References

[1] ZOU Z，SHI Z，GUO Y，et al.Object detection in 20 years：a survey[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2020，84（1）：154-191.
[2] SYED S A Z，MOHAMMAD S A，ASRA A，et al.A survey of modern deep learning based object detection models[J].IET Computer Vision，2021，24（4）：46-73.
[3] 王迪聪，白晨帅，邬开俊.基于深度学习的视频目标检测综述[J].计算机科学与探索，2021，15（9）：1563-1577.
WANG D C，BAI C S，WU K J.Survey of video object detection based on deep learning[J].Journal of Frontiers of Computer Science & Technology，2021，15（9）：1563-1577.
[4] 赵辉，李志伟，方禄发.特征信息增强的单发多框检测器算法[J].计算机工程与应用，2021，57（4）：148-154.
ZHAO H，LI Z W，FANG L F.Feature information enhancement based single shot multibox detector algorithm[J].Computer Engineering and Applications，2021，57（4）：148-154.
[5] LIN T，GOYAL P，GIRSHICK，et al.Focal loss for dense object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2017：2980-2988.
[6] TIAN Z，SHEN C.FCOS：fully convolutional one-stage object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2019：78-86.
[7] YANG Z，LIU S.RepPoints：point set representation for object detection[C]//IEEE International Conference on Computer Vision，2019：834-852.
[8] REN S，HE K，GIRSHICK R，et al.Faster R-CNN：towards real-time object detection with region proposal networks[C]//Advances in Neural Information Processing Systems，2015：91-99.
[9] 许德刚，王露，李凡.深度学习的典型目标检测算法研究综述[J].计算机工程与应用，2021，57（8）：10-25.
XU D G，WANG L，LI F.Review of typical object detection algorithms for deep learning[J].Computer Engineering and Applications，2021，57（8）：10-25.
[10] 罗会兰，彭珊，陈鸿坤.目标检测难点问题最新研究进展综述[J].计算机工程与应用，2021，57（5）：36-46.
LUO H L，PENG S，CHEN H K.Review on latest research progress of challenging problems in object detection[J].Computer Engineering and Applications，2021，57（5）：36-46.
[11] CAI Z，VASCONCELOS N.Cascade R-CNN：delving into high quality object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2018：6154-6162.
[12] LAW H，DENG J.CornerNet：detecting objects as paired keypoints[C]//Proceedings of the European Conference on Computer Vision，2018：754-770.
[13] 严春满，王铖.卷积神经网络模型发展及应用[J].计算机科学与探索，2021，15（1）：27-46.
YAN C M，WANG C.Development and application of convolutional neural network model[J].Journal of Frontiers of Computer Science & Technology，2021，15（1）：27-46.
[14] HE K，ZHANG X Y，REN S，et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2016：770-778.
[15] TAN M，PANG R.EfficientDet：scalable and efficient object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2020：384-402.
[16] GUO C，FAN B，ZHANG Q，et al.AugFPN：improving multi-scale feature learning for object detection[C]//IEEE Conference on Computer Vision and Pattern Recognition，2020：189-204.
[17] WANG T，ZHANG X，SUN J.Implicit feature pyramid network for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2021：117-134.
[18] 宋云博，陈冬艳，郝赟，等.基于级联卷积神经网络的高效目标检测方法[J].计算机工程与应用，2021，57（5）：139-145.
SONG Y B，CHEN D Y，HAO Y，et al.Efficient object detection method based on cascaded convolutional neural network[J].Computer Engineering and Applications，2021，57（5）：139-145.
[19] ZHOU X，WANG D.Bottom-up object detection by grouping extreme and center points[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2019：378-394.
[20] ZHOU X，WANG D.Objects as points[C]//Proceedings of the European Conference on Computer Vision，2018：326-341.
[21] ZHU C，HE Y.Feature selective anchor-free module for single-shot object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2019：462-481.
[22] ZHANG X，WAN F.FoveaBox：beyound anchor-based object detection[J].IEEE Transactions on Image Processing，2019，29（1）：7389-7398.
[23] TIAN Z，SHEN C.FCOS：a simple and strong anchor-free object detector[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2020：623-640.
[24] WANG N，GAO Y.NAS-FCOS：fast neural architecture search for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2020：216-232.
[25] ZHENG Z，WANG P.Distance-IoU loss：faster and better learning for bounding box regression[J].arXiv：1911. 08287，2019.
[26] 张朕通，单玉刚，袁杰.联合多尺度和注意力机制的遥感影像检测[J].计算机工程与应用，2021，57（9）：212-216.
ZHANG Z T，SHAN Y G，YUAN J.Remote sensing image detection algorithm combining multi-scale and attention mechanism[J].Computer Engineering and Applications，2021，57（9）：212-216.
[27] CAO Y，XU J，LIN S，et al.GCNet：non-local networks meet squeeze-excitation networks and beyond[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops，2019：435-451.
[28] LIU S，HUANG D.Receptive field block net for accurate and fast object detection[C]//European Conference on Computer Vision，2018.
[29] LIAO Y W，ZHANG G G，YANG Z G，et al.EFLDet：enhanced feature learning for object detection[J].Neural Computing and Applications，2022，34（2）：1033-1045.
[30] LIAO Y W，LU S W，YANG Z G，et al.Depthwise grouped convolution for object detection[J].Machine Vision and Applications，2021，32（6）：115.