eduzhai > Applied Sciences > Engineering >

Deep learning for lithological classification of carbonate rock micro-CT images

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 13 pages

Abstract: In addition to the ongoing development, pre-salt carbonate reservoircharacterization remains a challenge, primarily due to inherent geologicalparticularities. These challenges stimulate the use of well-establishedtechnologies, such as artificial intelligence algorithms, for imageclassification tasks. Therefore, this work intends to present an application ofdeep learning techniques to identify patterns in Brazilian pre-salt carbonaterock microtomographic images, thus making possible lithological classification.Four convolutional neural network models were proposed. The first modelincludes three convolutional layers followed by fully connected layers and isused as a base model for the following proposals. In the next two models, wereplace the max pooling layer with a spatial pyramid pooling and a globalaverage pooling layer. The last model uses a combination of spatial pyramidpooling followed by global average pooling in place of the last pooling layer.All models are compared using original images, when possible, as well asresized images. The dataset consists of 6,000 images from three differentclasses. The model performances were evaluated by each image individually, aswell as by the most frequently predicted class for each sample. According toaccuracy, Model 2 trained on resized images achieved the best results, reachingan average of 75.54 for the first evaluation approach and an average of 81.33 for the second. We developed a workflow to automate and accelerate thelithology classification of Brazilian pre-salt carbonate samples bycategorizing microtomographic images using deep learning algorithms in anon-destructive way.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×