Image Rotating Objects Detection Based on Single Level Feature Pyramid

doi:10.3778/j.issn.1002-8331.2301-0017

Abstract

Abstract: The object detection task of remote sensing image is a research hotspot in the field of remote sensing applications and has been widely concerned. However, with the improvement of the resolution of remote sensing images, the arbitrary orientation and the large aspect ratio of targets in remote sensing images are more obvious, which is a challenge for existing methods. In response to the above problems, this paper proposes a image rotation object detection model based on simple feature pyramid. Firstly, a simple feature pyramid structure is designed, and the multi-scale features of the target are obtained by combining the expansion convolution group. Secondly, the classification method is used to process the angle information of the rotating box, combined with the set prediction idea of DETR, a new bounding box regression loss is constructed to realize rotating targets detection without anchor frames. Finally, in order to reduce the amount of model computation and accelerates the convergence, weight constraints are added to the cross-attention of the decoder, which limit the global attention calculation to the local range. Experiments on the DOTA dataset show that this method not only improves the model detect performance, but also effectively solves the problem of slow convergence of DETR model.

Key words: remote sensing image, rotating target detection, single feature pyramid, cyclical focal loss, angle classification

摘要： 遥感图像的目标检测任务是遥感应用领域的一个研究热点，一直受到广泛的关注。随着遥感图像分辨率的提高，遥感图像中目标的多方向性、目标大纵横比等特点更加明显，这对于已有的方法是一个挑战。针对以上问题，提出了基于单级特征金字塔的图像旋转目标检测模型。设计了单级特征金字塔结构，并结合膨胀卷积组获得目标的多尺度特征；使用分类方法处理旋转框的角度信息，结合DETR的集合预测思想，构造新的边界框回归损失，实现无锚框旋转目标检测；为了减少模型计算量并加快收敛速度，在解码器的交叉注意力上加入权重约束，将全局注意力计算限制在局部范围内。在DOTA数据集上的实验证明，该方法在提升模型检测性能的同时，有效地解决了DETR模型收敛速度慢的问题。

关键词: 遥感图像, 旋转目标检测, 单级特征金字塔, 周期局部损失, 角度分类

ZHANG Zheng, BAI Jiahua, TIAN Qing. Image Rotating Objects Detection Based on Single Level Feature Pyramid[J]. Computer Engineering and Applications, 2023, 59(15): 235-242.

张正, 白佳华, 田青. 基于单级特征金字塔的图像旋转目标检测[J]. 计算机工程与应用, 2023, 59(15): 235-242.

References

[1] 李巍，戴朝霞，张向东，等.旋转目标检测算法在卫星影像中的应用[J].计算机工程与应用，2021，57（16）：134-141.
LI W，DAI Z X，ZHANG X D et al.Application of rotating target detection algorithm in satellite image[J].Computer Engineering and Applications，2021，57（16）：134-141
[2] 石颉，袁晨翔，丁飞，等.SAR图像建筑物目标检测研究综述[J].计算机工程与应用，2022，58（8）：58-66.
SHI J，YUAN C X，DING F，et al.Survey of building target detection in SAR images[J].Computer Engineering and Applications，2022，58（8）：58-66.
[3] ZHANG Y，HAO Y S.A survey of SAR image target detection based on convolutional neural networks[J].Remote Sensing，2022（14）：6240.
[4] 谭显东，彭辉.改进YOLOv5的SAR图像舰船目标检测[J].计算机工程与应用，2022，58（4）：247-254.
TAN X D，PENG H.Improved YOLOv5 ship target detection in SAR image[J].Computer Engineering and Applications，2022，58（4）：247-254.
[5] YAO L，LI M J，LI S S，et al.Improved YOLOv5 for remote sensing rotating object detection[C]//2021 6th International Conference on Communication，Image and Signal Processing（CCISP），2021：64-68.
[6] 张馨月，降爱莲.融合特征增强和自注意力的SSD小目标检测算法[J].计算机工程与应用，2022，58（5）：247-255.
ZHANG X Y，JIANG A L.SSD small target detection algorithm combining feature enhancement and self-attention[J].Computer Engineering and Applications，2022，58（5）：247-255.
[7] ASHISH V，SHAZEER N M，PARMAR N，et al.Attention is all you need[J].arXiv：1706.03762，2017.
[8] DING J，XUE N，LONG Y，et al.Learning RoI transformer for oriented object detection in aerial images[J]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition（CVPR），2019：2844-2853.
[9] XUE Y，LIU Q Q，YAN J C，et al.R3Det：refined single-stage detector with feature refinement for rotating object[C]//AAAI Conference on Artificial Intelligence，2019.
[10] 张哲晗，方薇，杜丽丽，等.基于编码-解码卷积神经网络的遥感图像语义分割[J].光学学报，2020，40（3）：46-55.
ZHANG Z H，FANG W，DU L L，et al.Semantic segmentation of remote sensing image based on encoder-decoder convolutional neural network[J].Acta Optica Sinica，2020，40（3）：46-55.
[11] HOU L P，LU K，XUE J，et al.Cascade detector with feature fusion for arbitrary-oriented objects in remote sensing images[C]//2020 IEEE International Conference on Multimedia and Expo（ICME），2020：1-6.
[12] XUE Y，YAN J C，MING Q，et al.Rethinking rotated object detection with gaussian wasserstein distance loss[C]//International Conference on Machine Learning，2021.
[13] LI Y Y，QIN H，XUAN P，et al.RADet：refine feature pyramid network and multi-layer attention network for arbitrary-oriented object detection of remote sensing images[J].Remote Sensing，2020（12）：389.
[14] XUE Y，YANG J R，YAN J C，et al.SCRDet：towards more robust detection for small，cluttered and rotated objects[C]//2019 IEEE/CVF International Conference on Computer Vision（ICCV），2018：8231-8240.
[15] MA T L，MAO M Y，ZHENG H H，et al.Oriented object detection with transformer[J].arXiv：2106.03146，2021.
[16] CARION N，MASSA F，SYNNAEVE G，et al.End-to-end object detection with transformers[J].arXiv：2005. 12872，2020.
[17] XUE Y，YAN J C.Arbitrary-oriented object detection with circular smooth label[C]//European Conference on Computer Vision，2020.
[18] LIN T Y，GOYAL P，GIRSHICK R B，et al.Focal loss for dense object detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2017，42：318-327.
[19] YU F，VLADLEN K.Multi-scale context aggregation by dilated convolutions[J].arXiv：1511.07122，2015.
[20] 刘腊梅，王晓娜，刘万军.融合转置卷积与深度残差图像语义分割方法[J].计算机科学与探索，2022，16（9）：2132-2142.
LIU L M，WANG X N，LIU W J，et al.Image semantic segmentation method with fusion of transposed convolution and deep residual[J].Jounal of Frontiers of Computer Science and Technology，2022，16（9）：2132-2142.
[21] ADELSON E H，BURT P J，ANDERSON C H，et al.Pyramid methods in image processing[J].RCA Engineer，1984.
[22] REN S Q，HE K M，GIRSHICK R B，et al.Faster R-CNN：towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2015，39：1137-1149.
[23] LIU W，ANGUELOV D，ERHAN D，et al.SSD：single shot multibox detector[C]//European Conference on Computer Vision，2015.
[24] LIN T Y，DOLLáR P，GIRSHICK R B，et al.Feature pyramid networks for object detection[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition（CVPR），2016：936-944.
[25] 欧阳柳，贺禧，瞿绍军.全卷积注意力机制神经网络的图像语义分割[J].计算机科学与探索，2022，16（5）：1136-1145.
OUYANG L，HE X，ZHAI S J.Fully convolutional neural network with attention module for semantic segmentation[J].Jounal of Frontiers of Computer Science and Technology，2022，16（5）：1136-1145.
[26] 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，2017：3974-3983.
[27] XU　Y　C，FU　M　T，WANG　Q　M，et al.Gliding vertex on the horizontal bounding box for multi-oriented object detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2021，43（4）：1452-1459.
[28] HAN J M，DING J，XUE N，et al.ReDet：a rotation-equivariant detector for aerial object detection[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition（CVPR），2021：2785-2794.
[29] XIE X X，CHENG G，WANG J B，et al.Oriented R-CNN for object detection[C]//2021 IEEE/CVF International Conference on Computer Vision（ICCV），2021：3500-3509.