Real-time precise object segmentation using a pixel-wise coarse-fine method with deep learning for automated manufacturing

被引:0
|
作者
Cho, Jaemin [1 ,2 ]
Kang, Sangseung [2 ]
Kim, Kyekyung [1 ,2 ]
机构
[1] Korea Univ Sci & Technol, Dept Comp Software, Daejeon 34113, South Korea
[2] Elect & Telecommun Res Inst ETRI, Daejeon 34129, South Korea
关键词
Manufacturing system; Computer vision; Convolutional neural network; Object detection; Object recognition; Precise segmentation; NETWORKS;
D O I
10.1016/j.jmsy.2021.11.004
中图分类号
T [工业技术];
学科分类号
08 ;
摘要
The Fourth Industrial Revolution has rapidly increased the demand, beyond the pace of industrial automation, for artificial intelligence (AI) and robots to undertake sophisticated tasks such as assembly and packaging without the use of feed devices. For commercial use, suitable technologies for stable real-time detection and precise segmentation are deemed essential. Particularly, because AI is used in an embedded environment, the model and algorithm should be lightweight. To satisfy the industrial requirements and solve related difficulties, this paper proposes a precision object segmentation method to accurately detect objects moving on a conveyor belt without using feed devices. A new backbone network was designed to reinforce the feature extraction layers using convolutional blocks, and it was added to the object detection component. To maintain lightweight compactness and complement feature mapping, a multi-level pooling layer was included. In addition, a loss function that focuses on difficult samples and a data augmentation method for learning were used to improve the performance without affecting the model architecture. In the segmentation step, the authors extracted and analyzed pixel-wise features to acquire more pertinent vertex candidates and utilized an intensity-difference search algorithm to select the optimal outline component among the candidates for precise segmentation. Finally, the posture information of the object was estimated using the object's geometric features to estimate the grasping point. The experimental results confirmed the improved precision and accuracy of the proposed method in object-detection and segmentation, demonstrating that the model is ideal for our manufacturing dataset.
引用
收藏
页码:114 / 123
页数:10
相关论文
共 50 条
  • [1] Real-Time and Deep Learning Based Vehicle Detection and Classification Using Pixel-Wise Code Exposure Measurements
    Kwan, Chiman
    Gribben, David
    Chou, Bryan
    Budavari, Bence
    Larkin, Jude
    Rangamani, Akshay
    Tran, Trac
    Zhang, Jack
    Etienne-Cummings, Ralph
    [J]. ELECTRONICS, 2020, 9 (06) : 1 - 21
  • [2] Robust Real-Time Visual Tracking Using Pixel-Wise Posteriors
    Bibby, Charles
    Reid, Ian
    [J]. COMPUTER VISION - ECCV 2008, PT II, PROCEEDINGS, 2008, 5303 : 831 - 844
  • [3] Pixel-wise real-time advanced calibration method for thermal infrared cameras
    Tremblay, Pierre
    Belhumeur, Louis
    Chamberland, Martin
    Villemaire, Andre
    Dubois, Patrick
    Marcotte, Frederic
    Belzile, Charles
    Farley, Vincent
    Lagueux, Philippe
    [J]. INFRARED IMAGING SYSTEMS: DESIGN, ANALYSIS, MODELING, AND TESTING XXI, 2010, 7662
  • [4] Civil infrastructure defect assessment using pixel-wise segmentation based on deep learning
    Savino, Pierclaudio
    Tondolo, Francesco
    [J]. JOURNAL OF CIVIL STRUCTURAL HEALTH MONITORING, 2023, 13 (01) : 35 - 48
  • [5] Civil infrastructure defect assessment using pixel-wise segmentation based on deep learning
    Pierclaudio Savino
    Francesco Tondolo
    [J]. Journal of Civil Structural Health Monitoring, 2023, 13 : 35 - 48
  • [6] Pixel-wise and real-time estimation of optical mean path length using deep learning: application for intraoperative functional brain mapping
    Caredda, Charly
    Ezhov, Ivan
    Sdika, Michael
    Lange, Frederic
    Giannoni, Luca
    Tachtsidis, Ilias
    Montcel, Bruno
    [J]. CLINICAL BIOPHOTONICS III, 2024, 13009
  • [7] Layer-wise anomaly detection and classification for powder bed additive manufacturing processes: A machine-agnostic algorithm for real-time pixel-wise semantic segmentation
    Scime, Luke
    Siddel, Derek
    Baird, Seth
    Paquit, Vincent
    [J]. ADDITIVE MANUFACTURING, 2020, 36
  • [8] A DEEP LEARNING APPROACH TO SPINE SEGMENTATION USING A FEED-FORWARD CHAIN OF PIXEL-WISE CONVOLUTIONAL NETWORKS
    Whitehead, William
    Moran, Steven
    Gaonkar, Bilwaj
    Macyszyn, Luke
    Iyer, Subramanian
    [J]. 2018 IEEE 15TH INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING (ISBI 2018), 2018, : 868 - 871
  • [9] Real-Time Automated Segmentation of Median Nerve in Dynamic Ultrasonography Using Deep Learning
    Yeh, Cheng-Liang
    Wu, Chueh-Hung
    Hsiao, Ming-Yen
    Kuo, Po-Ling
    [J]. ULTRASOUND IN MEDICINE AND BIOLOGY, 2023, 49 (05): : 1129 - 1136
  • [10] Detection of Subtle Bruises on Winter Jujube Using Hyperspectral Imaging With Pixel-Wise Deep Learning Method
    Feng, Lei
    Zhu, Susu
    Zhou, Lei
    Zhao, Yiying
    Bao, Yidan
    Zhang, Chu
    He, Yong
    [J]. IEEE ACCESS, 2019, 7 : 64494 - 64505