When automatically extracting fracture attitudes from a dense point cloud, there are challenges that complicate the correct characterization of the outcrop disrupted by certain fracture sets. In this regard, experimental and methodological studies were carried out to analyze fractures and their associated sets, which were extracted using an automated method to identify satisfactory conditions for data collection and processing. Primary data were acquired using a DJI Phantom 4 RTK unmanned aerial vehicle in the Primorsky and North Baikal deep fault zones of the Baikal rift. Structure from Motion photogrammetry was used to create dense point clouds. The obtained results show that the unidirectional camera orientation and the exposure of a single wall of the rock outcrop present a challenge for determining fracture sets. However, selecting a fragment of the digital outcrop model (DOM) facing various cardinal directions helps avoid orientation bias in the stereographic projections of poles to fractures and provides high-precision data on the fracture pattern. A well-chosen fragment of the DOM accurately represents fracture sets of the overall outcrop, saving time and computational resources. When the structural environment changes, the DOM allows for an analysis of the fracture pattern at any outcrop section, thereby improving the quality and productivity of the work. In the case of limited processing power, it is sufficient to use a dense point cloud of medium quality to extract fractures and their sets. The correct automated identification of fracture sets enhances the efficiency of rock deformation analysis in the field of engineering geology and mineral resource development.