Enhanced YOLOv8 for Drilling Core Image Segmentation

The exploration of deep formations by drilling has brought the issue of rock fragment into focus, emphasizing the need for accurate assessment of the fragment state. Core sampling, involving the extraction of cylindrical rock samples, has become a widely used method for evaluating formation properti...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Chinese Control Conference S. 7774 - 7779
Hauptverfasser: Yuan, Zhong, Duan, Longchen, Gao, Hui, Tan, Songheng
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: Technical Committee on Control Theory, Chinese Association of Automation 28.07.2025
Schlagworte:
Accuracy
Cascade Encoder-Decoder
Convolutional neural networks
Core Analysis
Deep learning
Drilling
Feature extraction
Image segmentation
Reliability
Rocks
Segmentation
Surface treatment
Switches
YOLOv8
ISSN:1934-1768
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The exploration of deep formations by drilling has brought the issue of rock fragment into focus, emphasizing the need for accurate assessment of the fragment state. Core sampling, involving the extraction of cylindrical rock samples, has become a widely used method for evaluating formation properties, including fragmentation. However, traditional core analysis is labor-intensive and challenges arise in accurately assessing fine fractured particles. Recent advancements in deep learning, particularly convolutional neural networks (CNN), have shown promise in automating the estimation of Rock Quality Designation (RQD), a key measure of rock mass quality. Despite these advancements, variations in core sizes and dense target distributions in core images pose significant challenges for segmentation. Existing research has focused primarily on larger intact cores, leaving a gap in the evaluation of smaller fractured rock fragments. This paper investigates the application of advanced segmentation techniques, including an enhanced version of YOLOv8, for core and fracture segmentation. The proposed model effectively addresses challenges associated with fine fractured particles and multi-scale feature extraction, improving segmentation accuracy and reliability. Results demonstrate that the enhanced YOLOv8 model outperforms traditional methods in extracting core fragments, offering a more scalable and efficient solution for fracture assessment in drilling operations.
ISSN:1934-1768
DOI:10.23919/CCC64809.2025.11178307