Anchor-Free Target Detection Model Based on Single Thread

doi:10.19678/j.issn.1000-3428.0057376

Abstract

Abstract: In order to avoid the negative impact of anchor on the robustness of the target detection model and ensure the model has high accuracy without anchor, this paper proposes a single thread and no anchor full convolution network model. By canceling the pre-set anchor parameters and pixel level prediction, the model has higher robustness when detecting targets without anchor. The hourglass backbone network is used to replace the feature pyramid module, which reduces the redundancy and calculation of anchor and feature pyramid module, and makes the overall model structure more simplified. The experimental results show that compared with the classical Retinanet model, the proposed model can significantly improve the prediction ability by using the positive case region principle and the center deviation branch, and has higher positive and negative label ratio and faster reasoning speed.

Key words: object detection, anchor-free network, single route, robustness of the model, pixel level prediction

摘要： 为避免锚点对目标检测模型的鲁棒性造成负面影响，并保证在无锚点情况下模型具有较高的准确度，提出一种单线程无锚点全卷积网络模型。通过取消预设锚点参数以及像素级别预测，使得模型在无锚点情况下检测目标时具有更高的鲁棒性。使用沙漏骨干网络取代特征金字塔模块，从而降低锚点与特征金字塔模块的冗余以及计算量，使整体模型结构更加精简。实验结果表明，与经典锚点Retinanet模型相比，该模型利用正例区域原则与中心偏离支路显著提高了预测能力，并且具有更高的正负标签比例和更快的推理速度。

关键词: 目标检测, 无锚点网络, 单线程, 模型鲁棒性, 像素级预测

CLC Number:

TP18

LI Hao, ZHANG Xiaoqiang. Anchor-Free Target Detection Model Based on Single Thread[J]. Computer Engineering, 2021, 47(5): 229-235,243.

李浩, 张晓强. 基于单线程的无锚点目标检测模型[J]. 计算机工程, 2021, 47(5): 229-235,243.

/ / Recommend / Download Citations

URL: http://www.ecice06.com/EN/10.19678/j.issn.1000-3428.0057376

http://www.ecice06.com/EN/Y2021/V47/I5/229

References

[1] BIAN Zhaofeng,ZHANG Xuegong.Pattern recognition[M].Beijing:Tsinghua University Press,2000.(in Chinese)边肇琪,张学工.模式识别[M].北京:清华大学出版社,2000.
[2] BENGIO Y.Learning deep architectures for AI[J].Foundations and Trends in Machine Learning,2009,2(1):1-55.
[3] YU Kai,JIA Lei,CHEN Yuqiang.Yesterday,today and tomorrow of deep learning[J].Computer Research and Development.2013,50(9):1799-1804.(in Chinese)余凯,贾磊,陈雨强,等.深度学习的昨天、今天和明天[J].计算机研究与发展,2013,50(9):1799-1804.
[4] TSUNG-YI L,PRIYA G,ROSS G,et al.Focal loss for dense object detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2020,42(1):318-327.
[5] WEI L,DRAGOMIR A,DUMITRU E,et al.SSD:single shot multibox detector[EB/OL].[2020-01-10].https://arxiv.org/pdf/1512.02325.pdf.
[6] JOESPH R,ALI F.YOLOv3:an incremental improvement[EB/OL].[2020-01-10].https://arxiv.org/abs/arXiv:1804.02767.
[7] ALEXER B,WANG C Y,HONG-YUAN M L.YOLOv4:optimal speed and accuracy of object detection[EB/OL].[2020-01-10].https://arxiv.org/abs/arXiv:2004.10934.
[8] REN S Q,HE K M,ROSS G,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.
[9] TIAN Zhi,SHEN Chunhua,CHEN Hao,et al.FCOS:fully convolutional one-stage object detection[EB/OL].[2020-01-10].https://arxiv.org/abs/arXiv:1904.01355.
[10] LAW H,DENG J.CornerNet:detecting objects as paired keypoints[EB/OL].[2020-01-10].https://arxiv.org/abs/arXiv:1808.01244.
[11] DUAN Kaiwen,BAI Song,XIE Lingxi,et al.CenterNet:keypoint triplets for object detection[EB/OL].[2020-01-10].https://arxiv.org/abs/arXiv:1904.08189.
[12] TSUNG-YI L,PIOTR D,ROSS G,et al.Feature pyramid networks for object detection[EB/OL].[2020-01-10].https://arxiv.org/pdf/1612.03144.pdf.
[13] JONATHAN L,EVAN S,TREVOR D.Fully convolutional networks for semantic segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(4):1-8.
[14] TIAN Zhi,HE Tong,SHEN Chunhua,et al.Decoders matter for semantic segmentation:data-dependent decoding enables flexible feature aggregation[C]//Proceedings of 2019 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2019:1023-1028.
[15] HE Tong,SHNE Chunhua,TIAN Zhi,et al.Knowledge adaptation for efficient semantic segmentation[C]//Proceedings of 2019 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2019:1039-1046.
[16] YIN Wei,LIU Yifan,SHEN Chunhua,et al.Enforcing geometric constraints of virtual normal for depth prediction[C]//Proceedings of 2019 IEEE/CVF International Conference on Computer Vision.Washington D.C.,USA:IEEE Press,2019:1088-1096.
[17] CAO Z,STOMAS S,SHIH-EN W,et al.Realtime multi-person 2D pose estimation using part affinity fields[EB/OL].[2020-01-10].https://arxiv.org/abs/arXiv:1611.08050.
[18] WANG Xinlong,ZHANG Rufeng,KONG Tao,et al.SOLOv2:dynamic and fast instance segmentation[C]//Proceedings of Advances in Neural Information Processing Systems[2020-03-23].https://arxiv.org/abs/arXiv:2003. 10152.
[19] ALEXANGER K,HE K M,ROSS G,et al.Panoptic segmentation[EB/OL].[2020-01-10].https://arxiv.org/abs/arXiv:1801.00868.
[20] HE Kaiming,ZHANG Xiangyu,REN Shaoqing,et al.Deep residual learning for image recognition[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2016:1-47.
[21] HUANG G,LIU Z,LAURENS M,et al.Densely connected convolutional networks[EB/OL].[2020-01-10].https://arxiv.org/abs/arXiv:1608.06993.

Please choose a citation manager

Content to export