A Deep Learning Model for Quick and Accurate Rock Recognition with Smartphones

Table 4

The presented method is compared with traditional recognition methods and feature extraction methods based on the rock thin section.

Recognition methods	Recognition period	Recognition process and cost estimate	Whether it can be applied in the field	Recognition accuracy
The presented method	Real-time recognition	Use a smartphone to take pictures of the rocks directly. Approximately $300	Can be recognized in the field	Meet the accuracy of the survey
Traditional recognition methods	More than 2 days	Collect rock specimens, make rock thin sections, and observe the thin sections under the microscope by professionals. Costs over $850	Unable to recognize in the field	Meet the accuracy of the survey
The method for analyzing and feature extraction based on the rock thin section	More than 2 days	Collect rock samples, make rock thin sections, and recognize in PC. Costs over $850	Unable to recognize in the field	Meet the accuracy of the survey