Scholarlink

<p>Unmanned aerial vehicle (UAV) imagery and deep learning-based approaches offer effective solutions for accurately identifying rocks and performing volume and surface area analyses because traditional methods are time-consuming and costly. This study aims to automatically detect rock blocks in a region and calculate their volumes and 3D surface areas. U-Net, Mask R-CNN, and YOLOv8n-seg methods were applied for the detection of rock blocks. Mask R-CNN and YOLOv8n-seg models were found unsuitable for this study due to overlapping issues and false positive detections. Considering these limitations, U-Net with DenseNet121 transfer learning method was preferred as it provided more balanced and reliable results. To further assess the robustness and generalization capability of the proposed model, a three-fold cross-validation experiment was conducted, yielding an average validation IoU of 0.85 and a training IoU of 0.84. The trained model was applied to the study area, and 3111 rock blocks were detected. The volumes and 3D surface areas of the rock blocks were calculated with a Python script as 540.13 m<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(^3\)</EquationSource> </InlineEquation> and 4284.42 m<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(^2\)</EquationSource> </InlineEquation>, respectively. This study demonstrates a fast, cost-effective, and reliable approach for rock segmentation and volume/surface area calculations, providing direct contributions to engineering applications through the determination of physical rock properties. In addition, the codes used in this study can automatically detect different geological formations from ortophotos. Also, volume and 3D surface area algorithms developed in this study can be used to calculate different types of objects.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A fast method for detecting rock blocks deposits and calculating volumes and 3D surface areas

  • 期刊论文

  • 发表日期：

    2026年4月13日

查看全文
  • Ali Polat

摘要

Unmanned aerial vehicle (UAV) imagery and deep learning-based approaches offer effective solutions for accurately identifying rocks and performing volume and surface area analyses because traditional methods are time-consuming and costly. This study aims to automatically detect rock blocks in a region and calculate their volumes and 3D surface areas. U-Net, Mask R-CNN, and YOLOv8n-seg methods were applied for the detection of rock blocks. Mask R-CNN and YOLOv8n-seg models were found unsuitable for this study due to overlapping issues and false positive detections. Considering these limitations, U-Net with DenseNet121 transfer learning method was preferred as it provided more balanced and reliable results. To further assess the robustness and generalization capability of the proposed model, a three-fold cross-validation experiment was conducted, yielding an average validation IoU of 0.85 and a training IoU of 0.84. The trained model was applied to the study area, and 3111 rock blocks were detected. The volumes and 3D surface areas of the rock blocks were calculated with a Python script as 540.13 m \(^3\) and 4284.42 m \(^2\) , respectively. This study demonstrates a fast, cost-effective, and reliable approach for rock segmentation and volume/surface area calculations, providing direct contributions to engineering applications through the determination of physical rock properties. In addition, the codes used in this study can automatically detect different geological formations from ortophotos. Also, volume and 3D surface area algorithms developed in this study can be used to calculate different types of objects.