基于多向特征金字塔的轻量级目标检测算法

白创; 王英杰; 闫昱; DJUKANOVIC Milena

doi:10.37188/CJLCD.2021-0018

您当前的位置：

首页 >

文章列表页 >

基于多向特征金字塔的轻量级目标检测算法

目标检测与跟踪 | 更新时间：2021-11-08

- 基于多向特征金字塔的轻量级目标检测算法
- Lightweight object detection algorithm based on multi-directional feature pyramid
- 液晶与显示 2021年36卷第11期页码：1516-1524
- 作者机构：
  
  1.长沙理工大学物理与电子科学学院, 湖南长沙 410114
  2.柔性电子材料基因工程湖南省重点实验室, 湖南长沙 410114
  3.黑山大学电气工程学院, 黑山共和国波德戈里察 81000
- 作者简介：
  
  [ "白创(1983-)，男，博士，讲师，2015年于华中科技大学获得博士学位，主要从事机器视觉与模式识别技术、深度神经网络学习算法方面的研究。E-mail：154317586@qq.com" ]
  [ "王英杰(1994-)，女，硕士研究生，2017年于天津师范大学获得学士学位，主要从事深度学习、机器视觉方面的研究。E-mail：1051815150@qq.com" ]
- 基金信息：
  
  中国-黑山科技合作委员会第三届例会交流合作项目(3-7);长沙理工大学“双一流”科学研究国际合作拓展项目(2019ic18);柔性电子材料基因工程湖南省重点实验室开放基金(202005)
- DOI：10.37188/CJLCD.2021-0018
  中图分类号： TP391.41
- 收稿日期：2021-01-23，
  
  修回日期：2021-03-28，
  
  纸质出版日期：2021-11
- 稿件说明：
移动端阅览
白创, 王英杰, 闫昱, 等. 基于多向特征金字塔的轻量级目标检测算法[J]. 液晶与显示, 2021,36(11):1516-1524.

Chuang BAI, Ying-jie WANG, Yu YAN, et al. Lightweight object detection algorithm based on multi-directional feature pyramid[J]. Chinese journal of liquid crystals and displays, 2021, 36(11): 1516-1524.
白创, 王英杰, 闫昱, 等. 基于多向特征金字塔的轻量级目标检测算法[J]. 液晶与显示, 2021,36(11):1516-1524. DOI： 10.37188/CJLCD.2021-0018.

Chuang BAI, Ying-jie WANG, Yu YAN, et al. Lightweight object detection algorithm based on multi-directional feature pyramid[J]. Chinese journal of liquid crystals and displays, 2021, 36(11): 1516-1524. DOI： 10.37188/CJLCD.2021-0018.

摘要

针对实时目标检测算法Tiny YOLOv3存在的深层特征难以提取、实时性不能满足需求、边界框定位不够准确的问题，提出一种改进的轻量级检测模型MTYOLO (MdFPN Tiny YOLOv3)。该模型构造多向特征金字塔网络(MdFPN)代替简单级联，充分完成了多层语义信息的提取与融合；使用深度可分离卷积代替标准卷积，有效降低网络复杂度，提升了检测的实时性能；采用Complete IOU loss (CIOU loss)代替MSE作为回归损失函数，极大地提高了边界框的回归精度。在PASCAL VOC和COCO数据集上对MTYOLO进行测试，结果表明，改进后模型的mAP最高可达到83.7%，检测速度最快可达到205 fps。

Abstract

Aiming at the problems that the deep features are difficult to extract

the real-time performance cannot meet the requirements

and the bounding box positioning is not accurate enough of the real-time object detection algorithm Tiny YOLOv3

an improved lightweight detection model MTYOLO (MdFPN Tiny YOLOv3) is proposed. The model constructs multi-directional feature pyramid network (MdFPN) instead of simple concatenation to adequately complete the extraction and fusion of multi-layer semantic information. The deep separable convolution is used instead of standard convolution to effectively reduce the complexity of the network and improve the real-time performance of detection. The complete IOU loss (CIOU loss) is used instead of MSE as the regression loss function

which greatly improves the regression accuracy of bounding box. The results which MTYOLO is tested on PASCAL VOC and COCO datasets show that the mAP of the improved model can reach 78.7% and the detection speed can reach 205 fps.

关键词

Keywords

references

王春哲 , 安军社 , 姜秀杰 , 等 . 基于卷积神经网络的候选区域优化算法 . 中国光学 , 2019 . 12 ( 6 ): 1348 - 1361 . https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGA201906018.htm https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGA201906018.htm .

C Z WANG , J S AN , X J JIANG , 等 . Region proposal optimization algorithm based on convolutional neural networks . Chinese Optics , 2019 . 12 ( 6 ): 1348 - 1361 . https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGA201906018.htm https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGA201906018.htm .

MINAR M R, NAHER J. Recent advances in deep learning: an overview[J]. arXiv : 1807.08169, 2018.

GIRSHICK R, DONAHUE J, DARRELL T, et al . Rich feature hierarchies for accurate object detection and semantic segmentation[C]// Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition . Columbus: IEEE, 2014.

GIRSHICK R. Fast R-CNN[C]// Proceedings of 2015 IEEE International Conference on Computer Vision . Santiago: IEEE, 2015: 1440-1448.

S Q REN , K M HE , R GIRSHICK , 等 . Faster R-CNN: towards real-time object detection with region proposal networks . IEEE Transactions on Pattern Analysis and Machine Intelligence , 2017 . 39 ( 6 ): 1137 - 1149 . DOI: 10.1109/TPAMI.2016.2577031 http://doi.org/10.1109/TPAMI.2016.2577031 .

T Y LIN , P GOYAL , R GIRSHICK , 等 . Focal loss for dense object detection . IEEE Transactions on Pattern Analysis and Machine Intelligence , 2020 . 42 ( 2 ): 318 - 327 . DOI: 10.1109/TPAMI.2018.2858826 http://doi.org/10.1109/TPAMI.2018.2858826 .

LIU W, ANGUELOV D, ERHAN D, et al . SSD: single shot MultiBox detector[C]// Proceedings of the 14 th European Conference on Computer Vision . Amsterdam, The Netherlands: Springer, 2016: 21-37.

REDMON J, DIVVALA S, GIRSHICK R, et al . You only look once: unified, real-time object detection[C]// Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition . Las Vegas: IEEE, 2016: 779-788.

REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]// Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition . Honolulu: IEEE, 2017: 6517-6525.

REDMON J, FARHADI A. YOLOv3: an incremental improvement[J]. arXiv : 1804.0276, 2018.

LIN T Y, DOLLÁR P, GIRSHICK R, et al . Feature pyramid networks for object detection[C]// Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition . Honolulu: IEEE, 2017.

LUO J H, WU J X, LIN W Y. ThiNet: a filter level pruning method for deep neural network compression[C]// Proceedings of 2017 IEEE International Conference on Computer Vision . Venice: IEEE, 2017.

马立 , 巩笑天 , 欧阳航空 . Tiny YOLOv3目标检测改进 . 光学精密工程 , 2020 . 28 ( 4 ): 988 - 995 . https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202004027.htm https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202004027.htm .

L MA , X T GONG , H K OUYANG . Improvement of Tiny YOLOV3 target detection . Optics and Precision Engineering , 2020 . 28 ( 4 ): 988 - 995 . https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202004027.htm https://www.cnki.com.cn/Article/CJFDTOTAL-GXJM202004027.htm .

W FANG , L WANG , P M REN . Tinier-YOLO: a real-time object detection method for constrained environments . IEEE Access , 2019 . 8 1935 - 1944 . http://ieeexplore.ieee.org/document/8941141 http://ieeexplore.ieee.org/document/8941141 .

WONG A, FAMUORI M, SHAFIEE M J, et al . YOLO Nano: a highly compact you only look once convolutional neural network for object detection[J]. arXiv : 1910.01271, 2019.

BOCHKOVSKIY A, WANG C Y, LIAO H Y M. YOLOv4: optimal speed and accuracy of object detection[J]. arXiv : 2004.10934, 2020.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于YOLOv8s改进的自动驾驶目标检测

基于动态特征剔除与轻量化检测的视觉SLAM算法

基于改进YOLOv8n的液晶屏Mura缺陷检测

高分辨率遥感图像的目标检测