Blasting, Volume. 42, Issue 3, 106(2025)

Blasting Fragmentation Recognition Method based on YOLOv8 and Binocular Vision

HUANG Lei1, TAO Ming2, LIU Yu-long1, XU Yuan-quan2、*, and XIANG Gong-liang2
Author Affiliations
  • 1CGNPC Uranium Resources Co., LTD., Beijing 100142, China
  • 2School of Resources and Safety Engineering, Central South University, Changsha 410083, China
  • show less

    To address the challenges of low efficiency, insufficient accuracy, and interference from complex environments in mining blast fragmentation recognition, this paper proposes a novel blast fragmentation recognition method based on binocular vision. By constructing a YOLOv8 instance segmentation model, the post-blast rock contours are accurately extracted under complex lighting conditions. By combining binocular measurement technology with the principles of three-dimensional coordinate transformation and disparity calculation, the maximum size of the fragments is determined. An indoor experimental platform was established to verify the accuracy of fragmentation recognition and size calculation under different parameters. Furthermore, an intelligent recognition architecture for open-pit mine blast fragmentation was proposed, and an automatic fragmentation recognition and analysis system was developed. The results of indoor simulation tests indicate that a lower camera height helps improve the model's recognition accuracy. Although fragment contact slightly affects the recognition of individual targets, the overall accuracy remains unaffected, with the recognition accuracy of all fragments exceeding 85%. The recognition accuracy slightly decreases in dynamic environments. However, the size calculation accuracy for 80% of the fragments remains above 90%, and the overall error remains within an acceptable range, meeting the requirements for real-time monitoring and subsequent analysis in blast fragmentation. This method has been successfully applied at the Husab Mine in Namibia, utilizing Radio Frequency Identification (RFID) technology to obtain material source information. It enables dynamic monitoring, precise analysis, and comprehensive evaluation of the fragment size distribution (FSD) throughout the entire block, providing a novel technological approach for assessing the effectiveness of open-pit bench blasting.

    Tools

    Get Citation

    Copy Citation Text

    HUANG Lei, TAO Ming, LIU Yu-long, XU Yuan-quan, XIANG Gong-liang. Blasting Fragmentation Recognition Method based on YOLOv8 and Binocular Vision[J]. Blasting, 2025, 42(3): 106

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category:

    Received: Feb. 15, 2025

    Accepted: Sep. 18, 2025

    Published Online: Sep. 18, 2025

    The Author Email: XU Yuan-quan (225511021@csu.edu.cn)

    DOI:10.3963/j.issn.1001-487x.2025.03.012

    Topics