Research on Intelligent Identification of Rock Types Based on Faster R-CNN Method

Xiaobo Liu,Huaiyuan Wang,Hongdi Jing,Liancheng Wang,Anlin Shao

doi:10.1109/access.2020.2968515

Xiaobo Liu, Huaiyuan Wang + Show 3 more

Open Access

https://doi.org/10.1109/access.2020.2968515

Copy DOI

Journal: IEEE Access	Publication Date: Jan 1, 2020
Citations: 50	License type: CC BY 4.0

Affiliation: Northeastern University, Ansteel (China)

Abstract

In the mining process of underground metal mines, the misjudgment of rock types by on-site technicians will have a serious negative impact on the stability evaluation of rock mass and the formulation of support schemes, which will result in the loss of economic benefits and potential safety hazards of mining enterprises. In order to realize the precise and intelligent identification of rock types, the image data of peridotite, basalt, marble, gneiss, conglomerate, limestone, granite, magnetite quartzite are amplified. Under the target detection framework of Faster R-CNN deep learning, the extraction network based on simplified VGG16 is used to extract and learn features of rock images, and finally the rock type identification system is successfully trained. The experimental verification shows that the system is correct for single-type rock image recognition and the accuracy is more than 96%. In order to realize accurate and intelligent identification of the surrounding rock surface under complex lithological conditions, the multi-type rocks hybrid images are also identified. The results show that the recognition effect is great and the accuracy rate is over 80%. Therefore, this system can accurately identify rock types with similar image features, which proves that the model has strong robustness and generalization ability. It has broad application prospects in rock mass stability evaluation and rock classification in underground mining.

Highlights

With the continuous development of human society and economy, countries around the world are paying increasingly attention to infrastructure construction
The results show that the simplified VGG16 convolutional neural network is highly applicable to rock type identification
The training data can be obtained by data amplification technology, and the GPU can be used to accelerate the training of the rock type identification model

Summary

INTRODUCTION

With the continuous development of human society and economy, countries around the world are paying increasingly attention to infrastructure construction. X. Liu et al.: Research on Intelligent Identification of Rock Types Based on Faster R-CNN Method. Cheng et al used image processing method to extract and train the gray image of rock slices, and obtained the rock type classification identification model. In the framework of Faster R-CNN deep learning target detection, the research used simplified VGG16 as the image feature extraction network, adjusted the RPN layer parameter settings, and trained different kinds of rock images through deep learning. CONSTRUCTION OF THE OVERALL FRAMEWORK STRUCTURE Faster R-CNN is mainly composed of RPN and Fast R-CNN [16] They share a simplified VGG16 network as the underlying feature extraction network. Liu et al.: Research on Intelligent Identification of Rock Types Based on Faster R-CNN Method TABLE 1. Based on the fixed shared convolutional layers, the Fast R-CNN was used to initialize the RPN network, and the RPN output candidate box was used as an input to adjust the Fast R-CNN parameters

Ncls i

Findings

CONCLUSION