[1]张 震,葛帅兵,陈可鑫,等.基于改进 YOLOv8 的遗留物品检测算法[J].郑州大学学报(工学版),2025,46(04):40-46.[doi:10.13705/j.issn.1671-6833.2025.04.010]
　ZHANG Zhen,GE Shuaibing,CHEN Kexin,et al.Abandoned Object Detection Algorithm Based on Improved YOLOv8[J].Journal of Zhengzhou University (Engineering Science),2025,46(04):40-46.[doi:10.13705/j.issn.1671-6833.2025.04.010]
点击复制

基于改进 YOLOv8 的遗留物品检测算法()
分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
46
期数:
2025年04期
页码:
40-46
栏目:
出版日期:
2025-07-10

文章信息/Info

Title:
Abandoned Object Detection Algorithm Based on Improved YOLOv8
文章编号:
1671-6833(2025)04-0040-07
作者:
张 震1;  葛帅兵2;  陈可鑫3;  李友好4;  黄伟涛4
1. 郑州大学 电气与信息工程学院,河南 郑州 450001;2. 郑州大学 河南先进技术研究院,河南 郑州 450003;3. 珠海优特电力科技股份有限公司,广东 珠海 519000;4. 河南汇融油气技术有限公司,河南 郑州 450001)
Author(s):
ZHANG Zhen1;  GE Shuaibing2;  CHEN Kexin3;  LI Youhao4;  HUANG Weitao4
1. School of Electrical and Information Engineering, Zhengzhou University, Zhengzhou 450001, China; 2. Henan Institute of Advanced Technology, Zhengzhou University, Zhengzhou 450003, China; 3. Zhuhai Unitech Power Technology Co. , Ltd. , Zhuhai 519000, China; 4. Henan Huirong Oil and Gas Technology Co. , Ltd. , Zhengzhou 450001, China
关键词:
遗留物品检测;  YOLOv8 算法;  EMA 注意力机制;  DySample 模块;  ADown 模块
Keywords:
abandoned object detection;  YOLOv8 algorithm;  EMA attention mechanisms;  DySample module;  ADown module
分类号:
TP391
DOI:
10.13705/j.issn.1671-6833.2025.04.010
文献标志码:
A
摘要:
针对传统基于背景减法的遗留物品检测算法难以应对人流拥挤、小目标、物品遮挡和光线变化等环境,以及基于深度学习方法中的模型准确率低等问题,提出了一种基于改进 YOLOv8 的遗留物品检测算法。 首先,使用动态上采样 DySample 替换最近邻上采样,优化上采样过程,增强模型的泛化能力。 其次,将高效轻量的 ADown 下采样模块替代普通的下采样卷积,在降低整个模型参数量的同时,提升算法的检测精度。 最后,引入 EMA 注意力机制,优化特征提取过程,增强特征提取能力,提升对小目标检测的效果。 实验结果表明:改进后的模型 YOLO-DAE 在自建数据集上取得的准确率 P、召回率 R、mAP@ 50 和 mAP@ 50:95 分别为 93. 4%,87. 7%,91. 7%和 80. 2%,相比于改进前的 YOLOv8s 模型在模型参数量和计算量减少的同时,分别提高了 1. 8 百分点、1. 6 百分点、1. 2 百分点和 2. 1 百分点,并且 mAP@ 50 和 mAP@ 50:95 均高于 YOLOv5s r6. 0、YOLOv6s v3. 0、YOLOv7s AF 和 YOLOv9s,有效提升了遗留物品检测能力。
Abstract:
An abandoned object detection algorithm based on improved YOLOv8 was proposed to address the difficulties of traditional background subtraction based abandoned object detection algorithms in dealing with crowded environments, small targets, occlusion, and light changes, as well as the low accuracy of models based on deep learning methods. Firstly, dynamic upsampling DySample was used to replace the nearest neighbor upsampling, optimizing the upsampling process, and increasing the model′s generalization ability. Secondly, the downsampling convolution was replaced with the efficient lightweight ADown module which reduced the overall model parameters while improving the detection accuracy of the algorithm. In addition, the introduction of EMA attention mechanism optimized the feature extraction process, enhanced feature extraction capabilities, and improved the effectiveness of small object detection. The experimental results showed that the improved model YOLO-DAE achieved P, R, and mAP@ 50 and mAP@ 50:95 was 93. 4%, 87. 7%, 91. 7%, and 80. 2%, respectively, which was 1. 8, 1. 6, 1. 2, and 2. 1 percentage points higher than the original YOLOv8s. And the average accuracy mAP@ 50 and mAP@ 50: 95 was higher than YOLOv5s r6. 0, YOLOv6s v3. 0, YOLOv7s AF, and YOLOv9s, effectively improving the ability to detect abandoned object。

参考文献/References:

[1] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation [ C ] ∥ 2014 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2014: 580-587. 

[2] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection [ C]∥2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway: IEEE, 2016: 779-788. 
[3] LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]∥14th European Conference on Computer Vision. Cham: Springer, 2016: 21-37. 
[4] TIAN Z, SHEN C H, CHEN H, et al. FCOS: fully convolutional one-stage object detection [ C] ∥2019 IEEE / CVF International Conference on Computer Vision ( ICCV) . Piscataway: IEEE, 2019: 9626-9635. 
[5] LIN K, CHEN S C, CHEN C S, et al. Abandoned object detection via temporal consistency modeling and backtracing verification for visual surveillance [ J ] . IEEE Transactions on Information Forensics and Security, 2015, 10(7) : 1359-1370. 
[6] SU H, WANG W, WANG S W. A robust all-weather abandoned objects detection algorithm based on dual background and gradient operator [ J] . Multimedia Tools and Applications, 2023, 82(19) : 29477-29499. 
[7] RUSSEL N S, SELVARAJ A. Ownership of abandoned object detection by integrating carried object recognition and context sensing[ J] . The Visual Computer, 2024, 40 (6) : 4401-4426. 
[8] PARK H, PARK S, JOO Y. Detection of abandoned and stolen objects based on dual background model and mask R-CNN[ J] . IEEE Access, 2020, 8: 80010-80019. 
[9] LIU W P, LIU P, XIAO C X, et al. General-purpose abandoned object detection method without background modeling [ C] ∥2021 IEEE International Conference on Imaging Systems and Techniques ( IST ) . Piscataway: IEEE, 2021: 1-5.
[10] 蒋晓可. 工业监控视频中的遗留物品检测与识别方法 研究[D] . 武汉: 华中科技大学, 2021. 
JIANG X K. Research on method of abandoned object detection and recognition in industrial surveillance videos [D] . Wuhan: Huazhong University of Science and Technology, 2021. 
[11] 林德钰, 周卓彤, 过斌, 等. 高斯混合模型与 GhostNet 结合的 YOLO-G 遗留物检测方法[ J] . 计算机辅助设 计与图形学学报, 2023, 35(1) : 99-107. 
LIN D Y, ZHOU Z T, GUO B, et al. YOLO-G abandoned object detection method combined with Gaussian mixture model and GhostNet[ J]. Journal of Computer-Aided Design & Computer Graphics, 2023, 35(1): 99-107. 
[12] Ultralytics. YOLOv8[EB / OL] . (2023- 01- 10) [ 2024- 09-10] . https:∥github. com / ultralytics/ ultralytics. 
[13] Ultralytics. YOLOv5[EB / OL] . (2020- 05- 18) [ 2024- 09-10] . https:∥github. com / ultralytics/ yolov5. 
[14] FENG C J, ZHONG Y J, GAO Y, et al. TOOD: taskaligned one-stage object detection[ C]∥2021 IEEE / CVF International Conference on Computer Vision ( ICCV ) . Piscataway: IEEE, 2021: 3490-3499. 
[15] LI X, WANG W H, WU L J, et al. Generalized focal loss: learning qualified and distributed bounding boxes for dense object detection[EB / OL]. (2020- 06- 08) [ 2024- 09-10] . https:∥arxiv. org / abs/ 2006. 04388. 
[16] LIU W Z, LU H, FU H T, et al. Learning to upsample by learning to sample[C]∥2023 IEEE / CVF InternationalConference on Computer Vision ( ICCV ) . Piscataway: IEEE, 2023: 6004-6014. 
[17] WANG C Y, YEH I H, LIAO H Y M. YOLOv9: learning what you want to learn using programmable gradient information[EB / OL] . ( 2024- 02- 29) [ 2024- 09- 10] . https:∥arxiv. org / abs/ 2402. 13616. 
[18] OUYANG D L, HE S, ZHANG G Z, et al. Efficient multi-scale attention module with cross-spatial learning [ C ] ∥ IEEE International Conference on Acoustics, Speech and Signal Processing ( ICASSP ) . Piscataway: IEEE, 2023: 1-5. 
[19] ZHANG Y F, SUN P Z, JIANG Y, et al. ByteTrack: multi-object tracking by associating every detection box [EB / OL] . (2022 - 04 - 07) [ 2024 - 09 - 10] . https:∥ arxiv. org / abs/ 2110. 06864. 
[20] SHI W Z, CABALLERO J, HUSZÁR F, et al. Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network [ C] ∥2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . Piscataway: IEEE, 2016: 1874-1883. 
[21] WANG J Q, CHEN K, XU R, et al. CARAFE: contentaware reassembly of features[C]∥2019 IEEE / CVF International Conference on Computer Vision ( ICCV) . Piscataway: IEEE, 2019: 3007-3016. 
[22] LU H, LIU W Z, FU H T, et al. FADE: fusing the assets of decoder and encoder for task-agnostic upsampling [EB / OL] . (2022 - 12 - 27) [ 2024 - 09 - 10] . https:∥ arxiv. org / abs/ 2207. 10392. 
[23] LU HAO, LIU W Z, YE Z X, et al. SAPA: similarityaware point affiliation for feature upsampling [ EB / OL] . (2022- 12 - 27 ) [ 2024 - 09 - 10 ] . https:∥arxiv. org / abs/ 2209. 12866. 
[24] HU J, SHEN L, ALBANIE S. Squeeze-and-excitation networks[ J] . IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020,42(8) :2011-2023. 
[25] WANG Q L, WU B G, ZHU P F, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]∥2020 IEEE / CVF Conference on Computer Vision and Pattern Recognition ( CVPR ) . Piscataway: IEEE, 2020: 11531-11539. 
[26] WOO S Y, PARK J C, LEE J Y, et al. CBAM: convolutional block attention module [ J ] . Lecture Notes in Computer Science, 2018, 11211: 3-19. 
[27] HUANG H J, CHEN Z G, ZOU Y, et al. Channel prior convolutional attention for medical image segmentation [EB / OL] . (2023 - 06 - 08) [ 2024 - 09 - 10] . https:∥ arxiv. org / abs/ 2306. 05196v1. 
[28] LI C Y, LI L L, GENG Y F, et al. YOLOv6 v3. 0: a fullscale reloading[ EB / OL]. ( 2023 - 01 - 13) [ 2024 - 09 - 10] . https:∥arxiv. org / abs/ 2301. 05586v1. 
[29] JIANG P T, ZHANG C B, HOU Q B, et al. LayerCAM: exploring hierarchical class activation maps for localization [ J] . IEEE Transactions on Image Processing, 2021, 30: 5875-5888.

更新日期/Last Update: 2025-07-13