In the worldwide challenge of climate change and growing demand for electricity, the need for eco-friendly and renewable energy sources has become more critical than ever before. Among various alternatives, photovoltaics has emerged as one of the most promising solutions. However, one of the key challenges of any large-scale PV installation is the frequent inspection required to maintain high operational efficiency. In this paper, we present an automated inspection method based on thermal imaging captured by Unmanned Aerial Vehicles. Our approach does not require any prior information about the photovoltaic farm, as it performs its reconstruction as a part of each inspection, making it robust to physical changes in the farm’s configuration. We successfully applied the proposed methodology to inspect solar farms at two locations in Poland during different seasons. This enabled us to track the progression of individual solar panels’ conditions over time. We conclude by demonstrating the effectiveness of our method through a comparison of defect detection results with expert annotations.