A multitude of applications in engineering, ore processing, mineral exploration, and environmental science require grain recognition and the counting of minerals. Typically, this task is performed manually with the drawback of monopolizing both time and resources. Moreover, it requires highly trained personnel with a wealth of knowledge and equipment, such as scanning electron microscopes and optical microscopes. Advances in machine learning and deep learning make it possible to envision the automation of many complex tasks in various fields of science at an accuracy equal to human performance, thereby, avoiding placing human resources into tedious and repetitive tasks, improving time efficiency, and lowering costs. Here, we develop deep-learning algorithms to automate the recognition of minerals directly from the grains captured from optical microscopes. Building upon our previous work and applying state-of-the-art technology, we modify a superpixel segmentation method to prepare data for the deep-learning algorithms. We compare two residual network architectures (ResNet 1 and ResNet 2) for the classification and identification processes. We achieve a validation accuracy of 90.5% using the ResNet 2 architecture with 47 layers. Our approach produces an effective application of deep learning to automate mineral recognition and counting from grains while also achieving a better recognition rate than reported thus far in the literature for this process and other well-known, deep-learning-based models, including AlexNet, GoogleNet, and LeNet.

The Opatica plutonic belt, Abitibi greenstone belt, and Pontiac Subprovince represent a major proportion of the southeastern Superior Province, which was formed and accreted rapidly between approximately 2.9 and 2.8 Ga. Plutons in these belts are grouped into four types: (i) trondhjemite–tonalite–granodiorite (TTG) suite (2.82–2.69 Ga), (ii) monzodiorite (MZD) suite (2.697–2.669 Ga), (iii) late alkaline granitoid (ALK) suite (2.68–2.67 Ga), and (iv) anatectic granite and monzonite (ANA) suite (2.69–2.64 Ga). The four suites are represented in all belts and show similar petrography and geochemistry. In terms of Nd-isotope composition, the TTG, MZD, and ALK suites are typical of destructive plate margin magmatism and have + 1.4 < initial ε_Nd < +3.7, values which are very similar to that of the Abitibi mantle (ε_Nd + 2.5). The lower values for the ANA suites (ε_Nd + 0.1 to + 2.4) result from recycling of crustal components. In the Opatica belt the ANA granitoids fall on the Nd-isotope evolution curve defined by the Opatica TTG plutons, and are thus considered to be melt products of this suite. However, Abitibi and Pontiac ANA suites show a larger range of ε_Nd, from + 0.1 to + 2.4, compared with + 1.0 to + 1.3 for the Opatica, suggesting more heterogeneous crustal source rock. Recent geological mapping and geophysical studies associated with the Lithoprobe project have suggested that the Opatica belt represents a plutonic belt against which the Abitibi was accreted by subduction-related collision and that the Pontiac Subprovince is dominated by imbricated metasediments related to the final stages of collision in the Abitibi region. The Nd-isotope data provide support for these arguments. Early plutonic suites are mantle derived and related to arc-accretion processes. As the collision process progresses, a more evolved isotopic component is introduced, possibly in relation to sediment subduction into the mantle. Anatexis of the crust in the central Opatica belt and the core of the Pontiac Subprovince resulted in the formation of granites with a crustal signature for Nd isotopes.