[1]蔡钟晟,陈 飞,曾勋勋.一种具有抗噪性能的圆形目标检测器[J].南京师大学报(自然科学版),2021,44(04):85-93.[doi:10.3969/j.issn.1001-4616.2021.04.011]
　Cai Zhongsheng,Chen Fei,Zeng Xunxun.A Circular Object Detector with Anti-Noise Performance[J].Journal of Nanjing Normal University(Natural Science Edition),2021,44(04):85-93.[doi:10.3969/j.issn.1001-4616.2021.04.011]
点击复制

一种具有抗噪性能的圆形目标检测器()
分享到：

《南京师大学报（自然科学版）》[ISSN:1001-4616/CN:32-1239/N]

卷:
第44卷
期数:
2021年04期
页码:
85-93
栏目:
·计算机科学与技术·
出版日期:
2021-12-15

文章信息/Info

Title:
A Circular Object Detector with Anti-Noise Performance
文章编号:
1001-4616(2021)04-0085-09
作者:
蔡钟晟; 陈 飞; 曾勋勋
福州大学数学与计算机科学学院,福建 福州 350108
Author(s):
Cai Zhongsheng; Chen Fei; Zeng Xunxun
College of Mathematics and Computer Science,Fuzhou University,Fuzhou 350108,China
关键词:
噪声图像; 目标检测; IOU损失函数; 形状特征; 先验信息
Keywords:
noisy image; object detection; IOU loss; shape feature; prior information
分类号:
TP751
DOI:
10.3969/j.issn.1001-4616.2021.04.011
文献标志码:
A
摘要:
目前许多目标检测模型在噪声环境中都会出现精度下降. 为了提高目标检测模型在噪声环境中的精度,本文从两方面提高检测精度. 首先,提出一个基于残差结构的抗噪特征提取模块,为后续的网络模块提供支撑. 其次,利用目标的先验信息,针对性地改进模型的锚框设计和损失函数设计. 根据目标的形状先验信息设计锚框的形状. 将IOU损失函数作为模型的Bbox损失函数,其中IOU损失项及最小闭包区域根据目标形状先验信息计算. 实验数据集为血细胞数据集和下旁腺数据集,基准对照模型为Yolov3和RetinaNet,同时也可移植到其他检测模型. 在血细胞数据集环境中,比较Yolov3的精度由62.7提高到75.7. 在下旁腺数据集中同样有所提升.
Abstract:
At present,many of the target detection models will suffer from precision degradation in noisy environment. In order to improve the accuracy of target detection model in noisy environment,two methods are proposed. Firstly,an anti-noise feature extraction module based on residual structure is proposed to support the subsequent network modules. Secondly,the anchor frame design and loss function design of the model are improved by using the prior information of the target. The shape of the anchor frame is designed according to the shape prior information of the target. The IOU loss function is taken as the Bbox loss function of the model,in which the IOU loss term and the minimum closure area are calculated according to the prior information of the target shape. The experimental data sets were blood cell data sets and inferior parathyroid gland data sets,and the benchmark control models were Yolov3 and RetinaNet,which can also be transplanted to other detection models. In the blood cell dataset environment,the accuracy of Yolov3 is 62.7,which is improved to 75.7,which is a great improvement. In the inferior parathyroid gland data set,there was also an improvement.

参考文献/References:

[1] ZHAO X,SATOH Y,TAKAUJI H,et al. Robust adapted object detection under complex environment[C]//2011 8th IEEE International Conference on Advanced Video and Signal Based Surveillance. Woshington DC:IEEE,2011:261-266.
[2]REDMON J,FARHADI A. Yolov3:an incremental improvement[C]//Computer Vision and Pattern Recognition. Guangzhou,2018.
[3]SUHAIL A,JAYABALAN M,THIRUCHELVAM V. Convolutional neural network based object detection:a review[J]. Journal of critical reviews,2020,7(11):2020.
[4]REN S,HE K,GIRSHICK R,et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. IEEE transactions on pattern analysis & machine intelligence,2017,39(6):1137-1149.
[5]张中宝,王洪元,张继,等. 基于Faster-RCNN的遥感图像飞机检测算法[J]. 南京师大学报(自然科学版),2018,41(4):79-86.
[6]HE K,GKIOXARI G,DOLLáR P,et al. Mask R-CNN[J]. IEEE transactions on pattern analysis & machine intelligence,2017,38(5):2961-2969.
[7]TIAN Z,SHEN C,CHEN H,et al. FCOS:fully convolutional one-stage object detection[C]//2019 IEEE/CVF International Conference on Computer Vision. Seanl,Korea:IEEE,2020.
[8]LAW H,DENG J. CornerNet:detecting objects as paired keypoints[J]. International journal of computer vision,2018,31(10):734-750.
[9]REDMON J,FARHADI A. YOLO9000:better,faster,stronger[J]. IEEE conference on computer vision and pattern recognition,2017:6517-6525.
[10]ZHAO Z Q,ZHENG P,XU S,et al. Object detection with deep learning:a review[J]. IEEE transactions on neural networks and learning systems,2019,30(11):3212-3232.
[11]WONG J A,HARTIGANM A. Algorithm AS 136:a K-means clustering algorithm[J]. Journal of the royal statistical society,1979,28(1):100-108.
[12]JIAN M,WANG J,YU H,et al. Visual saliency detection by integrating spatial position prior of object with background cues[J]. Expert systems with applications,2020,168(11):1142-1153.
[13]夏晨星. 基于背景先验和目标信息的自底向上显著性检测方法研究[D]. 长沙:湖南大学,2019.
[14]ZHAO M,CHE X,LIU H,et al. Medical prior knowledge guided automatic detection of coronary arteries calcified plaque with cardiac CT[J]. Electronics,2020,9(12):2122-2134.
[15]ZHAO H,ZHANG Z. Improving neural network detection accuracy of electric power bushings in infrared images by hough transform[J]. Sensors,2020,20(10):2931-2941.
[16]BALLARD D H. Generalizing the hough transform to detect arbitrary shapes[J]. Pattern recognition,1981,13(2):111-122.
[17]赵振兵,李延旭,甄珍,等. 结合KL散度和形状约束的Faster R-CNN典型金具检测方法[J]. 高电压技术,2020,46(9):3018-3026.
[18]REZATOFIGHI H,TSOI N,GWAK J Y,et al. Generalized intersection over union:a metric and a loss for bounding box regression[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Xi’an:IEEE,2019.
[19]ZHENG Z,WANG P,LIU W,et al. Distance-IoU loss:faster and better learning for bounding box regression[C]//AAAI Conference on Artificial Intelligence. New York,2020.
[20]HE K M,ZHANG X,REN S,et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,America:2016.
[21]ZHANG K,ZUO W,CHEN Y,et al. Beyond a gaussian denoiser:residual learning of deep CNN for image denoising[J]. IEEE transactions on image processing,2016,26(7):3142-3155.
[22]WANG J Q,CHEN K,YANG S,et al. Region proposal by guided anchoring[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Xi’an,2019.
[23]ZHANG S,CHI C,YAO Y,et al. Bridging the gap between anchor-based and anchor-free detection via adaptive training sample selection[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle,America:IEEE,2020.
[24]KONG T,SUN F,LIU H,et al. FoveaBox:beyound anchor-based object detection[J]. IEEE transactions on image processing,2020,29:7389-7398.
[25]PANG J,CHEN K,SHI J,et al. Libra R-CNN:towards balanced learning for object detection[C]//Conference on Computer Vision and Pattrn Recognition. Seattle,America,2020.
[26]LIN T Y,GOYAL P,GIRSHICK R,et al. Focal loss for dense object detection[J]. IEEE transactions on pattern analysis & machine intelligence,2020,42(2):318-327.
[27]LIU W,ANGUELOV D,ERHAN D,et al. SSD:single shot multibox detector[C]//European Conference on Computer Vision. Amsterdam Netherlands,2016.
[28]FERGUS R,TAYLOR G W,ZEILER M D. Adaptive deconvolutional networks for mid and high level feature learning[C]//International Conference on Computer Vision. Shenzhen:IEEE Computer Society,2011.

相似文献/References:

[1]郑 冬,李向群,许新征.基于轻量化SSD的车辆及行人检测网络[J].南京师大学报(自然科学版),2019,42(01):73.[doi:10.3969/j.issn.1001-4616.2019.01.012]
　Zheng Dong,Li Xiangqun,Xu Xinzheng.Vehicle and Pedestrian Detection Model Based on Lightweight SSD[J].Journal of Nanjing Normal University(Natural Science Edition),2019,42(04):73.[doi:10.3969/j.issn.1001-4616.2019.01.012]
[2]严 忱,严云洋,高尚兵,等.基于多级特征融合的视频火焰检测方法[J].南京师大学报(自然科学版),2021,44(03):131.[doi:10.3969/j.issn.1001-4616.2021.03.019]
　Yan Chen,Yan Yunyang,Gao Shangbing,et al.Video Flame Detection Based on Fusion of Multilevel Features[J].Journal of Nanjing Normal University(Natural Science Edition),2021,44(04):131.[doi:10.3969/j.issn.1001-4616.2021.03.019]

备注/Memo

备注/Memo:
收稿日期:2021-05-16.
基金项目:国家自然科学基金项目(61771141)、福建省教育厅中青年教师教育科研项目(JAT190020).
通讯作者:陈飞,博士,副教授,研究方向:图像处理与机器学习. E-mail:chenfei314@fzu.edu.cn
更新日期/Last Update: 2021-12-15