Аннотация:Core photo lithological interpretation based on computer analyses. Demands for speed which geological data should be processed is rising. The problem doubles when you don't have any unite Government or production standards for the processing of it. In addition to some geological tasks, there are no criteria for subjectivism reduction or hard to distinguish in boundary rock task. On the basis of machine learning and neural networks, we demonstrate the automatization of geological information analysis based on core images. We show the ability of the algorithm to distinguish six lithotypes with highest precision 98 % on several classes.