A mini dataset of lithology microscopic images is generated for both multi-class and binary classification tasks.
Lithology discrimination is a crucial activity in characterizing oil reservoirs, and processing lithology microscopic images is an essential technique for investigating fossils and minerals and geological assessment of shale oil exploration. In this study, a mini dataset of lithology microscopic images is generated for both multi-class and binary classification task. In the following, the process of producing the mini dataset is described in detail and step by step.
Core samples used in this dataset have been taken from 2905 meters to 3101 meters of an oil carbonate reservoir in the Southern Persian Gulf. Core studies provide an array of useful information on petrophysical features of the study area. The hydrocarbon formation studied comprises mainly oil and water without gas. More investigations show that the average porosity and permeability in the studied zone are X % and Y (millidarcy) MD, respectively. The study of thin sections shows that there are three main lithologies, including Argillaceous Limestone, Limestone, and Dolomite, as shown in Fig. 1. Argillaceous Limestone is a type of sedimentary rock that includes a significant amount (but less than 50%) of clay comprising kaolinite, montmorillonite, illite, and chlorite. Limestone is another sedimentary rock that is largely comprised of calcium carbonate (CaCO3), typically in the form of calcite or aragonite. Dolomite, widely recognized as Dolostone, is a sedimentary carbonate rock rich in CaMg (CO3)2. The lithology column effectively shows the intervals of rocks and how they are placed versus depth, which gives a better understanding of the studied area. In this study, we provided a lithology column after scrutinizing cores and microscopic images in the laboratory in the RB well. Regarding Fig. 1, Argillaceous Limestone is responsible for the most significant share of lithology; however, the contributions of Dolomite and Limestone are almost the same.
This repo contains the following items:
- Dataset Folder 'DS_Lithology_new':
- Dataset loader 'dataset_loader.py':
- An example 'DL_Lithology_microscopic_images_classification_task':
The folder 'DS_Lithology_new' is the Lithology-microscopic-images-mini-dataset and contains two sub-folders:
- The 'train_set': All train image samples are held in this sub-folder.
- The 'test_set': All test image samples are held in this sub-folder.
This function is provided to load and produce train and test sets.
In order to use the dataset, first you should download the folder 'DS_Lithology_new'. Then, use the 'dataset_loader.py' to load images (this function returns X_train, Y_train, X_test, Y_test).
This Dataset was developed under supervision of Dr. Keyvan RahimiZadeh and in collabotion with Prof. Amin Beheshti.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent