YOLO-v1 to YOLO-v8, the Rise of YOLO and Its Complementary Nature toward Digital Manufacturing and Industrial Defect Detection

被引:271
作者
Hussain, Muhammad [1 ]
机构
[1] Univ Huddersfield, Sch Comp & Engn, Dept Comp Sci, Queensgate, Huddersfield HD1 3DH, England
关键词
industrial defect detection; object detection; smart manufacturing; quality inspection; BATCH NORMALIZATION; DEEP; OPTIMIZATION; NETWORKS;
D O I
10.3390/machines11070677
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Since its inception in 2015, the YOLO (You Only Look Once) variant of object detectors has rapidly grown, with the latest release of YOLO-v8 in January 2023. YOLO variants are underpinned by the principle of real-time and high-classification performance, based on limited but efficient computational parameters. This principle has been found within the DNA of all YOLO variants with increasing intensity, as the variants evolve addressing the requirements of automated quality inspection within the industrial surface defect detection domain, such as the need for fast detection, high accuracy, and deployment onto constrained edge devices. This paper is the first to provide an in-depth review of the YOLO evolution from the original YOLO to the recent release (YOLO-v8) from the perspective of industrial manufacturing. The review explores the key architectural advancements proposed at each iteration, followed by examples of industrial deployment for surface defect detection endorsing its compatibility with industrial requirements.
引用
收藏
页数:25
相关论文
共 97 条
  • [21] Farahnakian F., 2021, P 2021 26 INT C AUT, DOI [10.23919/ICAC50006.2021.9594180, DOI 10.23919/ICAC50006.2021.9594180]
  • [22] Object Detection with Discriminatively Trained Part-Based Models
    Felzenszwalb, Pedro F.
    Girshick, Ross B.
    McAllester, David
    Ramanan, Deva
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2010, 32 (09) : 1627 - 1645
  • [23] Fu C.-Y., 2017, arXiv
  • [24] Theoretical analysis of skip connections and batch normalization from generalization and optimization perspectives
    Furusho, Yasutaka
    Ikeda, Kazushi
    [J]. APSIPA TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING, 2020, 9
  • [25] Dropout vs. batch normalization: an empirical study of their impact to deep learning
    Garbin, Christian
    Zhu, Xingquan
    Marques, Oge
    [J]. MULTIMEDIA TOOLS AND APPLICATIONS, 2020, 79 (19-20) : 12777 - 12815
  • [26] Ge Z, 2021, Arxiv, DOI [arXiv:2107.08430, 10.48550/arXiv.2107.08430, DOI 10.48550/ARXIV.2107.08430]
  • [27] Gevorgyan Z, 2022, Arxiv, DOI [arXiv:2205.12740, DOI 10.48550/ARXIV.2205.12740]
  • [28] Fast R-CNN
    Girshick, Ross
    [J]. 2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2015, : 1440 - 1448
  • [29] Region-Based Convolutional Networks for Accurate Object Detection and Segmentation
    Girshick, Ross
    Donahue, Jeff
    Darrell, Trevor
    Malik, Jitendra
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2016, 38 (01) : 142 - 158
  • [30] Recent advances in convolutional neural networks
    Gu, Jiuxiang
    Wang, Zhenhua
    Kuen, Jason
    Ma, Lianyang
    Shahroudy, Amir
    Shuai, Bing
    Liu, Ting
    Wang, Xingxing
    Wang, Gang
    Cai, Jianfei
    Chen, Tsuhan
    [J]. PATTERN RECOGNITION, 2018, 77 : 354 - 377