Table 5 SALM labels each component and counts the number of pixels.

From: Research on the generation and annotation method of thin section images of tight oil reservoir based on deep learning

 

Quantity and proportion of each component of the original image

Generate the number and proportion of each component of the image

Correctly label the quantity of each ingredient

Number and proportion of pixels in each component of the original image

Generate the number and proportion of pixels of each component of the image

Correctly label the number of pixels of each component

Quartz

2176 (14.07%)

1829 (11.49%)

1372

3,310,625 (25.64%)

3,207,254 (24.47%)

3,038,582

Feldspar

447 (2.89%)

426 (2.68%)

312

723,542 (5.60%)

692,335 (5.28%)

644,823

Lithic

8405 (63.40%)

10,862 (68.22%)

7433

7,320,953 (56.71%)

7,624,569 (58.17%)

6,759,211

PP

768 (4.97%)

847 (5.32%)

624

378,645 (3.01%)

394,852 (3.01%)

365,868

CP

144 (0.93%)

191 (1.20%)

138

133,529 (1.12%)

146,878 (1.12%)

135,249

CDP

2091 (13.51%)

1719 (10.80%)

1250

991,202 (7.68%)

989,052 (7.55%)

920,952

Microcrack

35 (0.23%)

46 (0.29%)

34

51,534 (0.40%)

52,538 (0.40%)

49,347