Automatic classification of volcanic rocks from thin section images using transfer learning networks

Polat, Özlem; Polat, Ali; Eki̇ci̇, Taner

doi:10.1007/s00521-021-05849-3

Cited by 35 publications

(11 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionally, Figure 26 shows GoogLeNet and VGG16's classification precision for the three types of rock with the use of Adam, SGD and RMSprop. It could be concluded that the classification effect of the model trained with the SGD optimizer was worse than that of the other two optimizers for both GoogLeNet and VGG16, which is also basically consistent with the conclusion of [30].…”

Section: Discussionsupporting

confidence: 81%

“…Additionally, Figure 26 shows GoogLeNet and VGG16's classification preci the three types of rock with the use of Adam, SGD and RMSprop. It could be con that the classification effect of the model trained with the SGD optimizer was wor that of the other two optimizers for both GoogLeNet and VGG16, which is also b consistent with the conclusion of [30]. In summary, the best options for the intelligent classification of rock thin-section images are the cosine decay mode, RMSprop optimizer, and VGG16 classification model.…”

Section: Discussionsupporting

confidence: 79%

See 1 more Smart Citation

Experimental Studies on Rock Thin-Section Image Classification by Deep Learning-Based Approaches

Zhang

2022

Mathematics

View full text Add to dashboard Cite

Experimental studies were carried out to analyze the impact of optimizers and learning rate on the performance of deep learning-based algorithms for rock thin-section image classification. A total of 2634 rock thin-section images including three rock types—metamorphic, sedimentary, and volcanic rocks—were acquired from an online open-source science data bank. Four CNNs using three different optimizer algorithms (Adam, SGD, RMSprop) under two learning-rate decay schedules (lambda and cosine decay modes) were trained and validated. Then, a systematic comparison was conducted based on the performance of the trained model. Precision, f1-scores, and confusion matrix were adopted as the evaluation indicators. Trials revealed that deep learning-based approaches for rock thin-section image classification were highly effective and stable. Meanwhile, the experimental results showed that the cosine learning-rate decay mode was the better option for learning-rate adjustment during the training process. In addition, the performance of the four neural networks was confirmed and ranked as VGG16, GoogLeNet, MobileNetV2, and ShuffleNetV2. In the last step, the influence of optimization algorithms was evaluated based on VGG16 and GoogLeNet, and the results demonstrated that the capabilities of the model using Adam and RMSprop optimizers were more robust than that of SGD. The experimental study in this paper provides important practical value for training a high-precision rock thin-section image classification model, which can also be transferred to other similar image classification tasks.

show abstract

Section: Discussionsupporting

confidence: 81%

Section: Discussionsupporting

confidence: 79%

Experimental Studies on Rock Thin-Section Image Classification by Deep Learning-Based Approaches

Zhang

2022

Mathematics

View full text Add to dashboard Cite

show abstract

“…However, conventional extraction techniques are unable to produce effective recognition results in thin mineral-rich sandstone sections. Thus, deep learning provides a new method for rock thin section identification . Borazjani et al developed two intelligent models to identify carbonate reservoir structures and air attack spaces, and achieved positive outcomes.…”

Section: Introductionmentioning

confidence: 99%

Multichannel Sandstone Thin Sections Identification Based on Improved DeepLab V3 Plus Neural Network

Zhong,

Meng,

Liu

2024

ACS Omega

View full text Add to dashboard Cite

“…Wanhyuk Seo [12] use ResNet152 to classify lithofacies thin sections and use Grad-CAM to visualize the output features of the last convolutional layer. Polat [13] transfer learned two CNN models to classify volcanic rocks. The above methods have achieved great results.…”

Section: Introductionmentioning

confidence: 99%

Petrographic Thin Section Image Classification Based on Deep Convolutional Network with Hybrid Attention Mechanism

Pan,

Han,

Guo

et al. 2023

Preprint

View full text Add to dashboard Cite

The classification of petrographic thin section image holds significant importance in the fields of geology and petroleum exploration. In light of the issue that the feature information of current petrographic thin section image recognition algorithms is prone to loss, this paper utilizes deep learning technologies and proposes the adoption of the Inception-ResNet-V2 network as the backbone network. When compared to various classical classification algorithms, the model proposed in this paper demonstrates a considerable advantage in classification accuracy on the petrographic thin section image dataset, achieving an impressive accuracy rate of 95.89%. The classification of each image category was analyzed by confusion matrix, and the model was visualized using Grad-CAM to observe the distribution of classification weights and the changes of classification weights. The results indicate that the model can extract crucial feature information from the petrographic thin section image, effectively enhancing the accuracy of petrographic thin section image recognition.

show abstract

Automatic classification of volcanic rocks from thin section images using transfer learning networks

Cited by 35 publications

References 27 publications

Experimental Studies on Rock Thin-Section Image Classification by Deep Learning-Based Approaches

Experimental Studies on Rock Thin-Section Image Classification by Deep Learning-Based Approaches

Multichannel Sandstone Thin Sections Identification Based on Improved DeepLab V3 Plus Neural Network

Petrographic Thin Section Image Classification Based on Deep Convolutional Network with Hybrid Attention Mechanism

Contact Info

Product

Resources

About