Abstract <p>The article deals with the problem of detecting potato tubers with phytodiseases and mechanical damage in the early stages, which is critically important to prevent spoilage of the harvested crop during storage and transportation. The authors explore the method of active thermal control as a promising alternative to traditional methods (visual, multispectral, hyperspectral). The main idea of the method is to periodically heat the tuber surface with infrared radiation and then analyze not only the amplitude but also the phase of temperature fluctuations on the surface using a thermal imaging camera. Numerical modeling in the COMSOL environment and subsequent experiments on real samples have shown that the proposed method makes it possible to effectively detect both surface and subsurface defects (for example, dry rot) at a depth of up to 2 mm, while reducing the influence of contamination, glare, and uneven illumination due to the curvature of the surface. It has been established that low periodic heating frequencies (less than 0.1 Hz) and a heat flux density of up to 1700 W/m² should be used for optimal detection of defects at various depths. The use of an artificial neural network to classify images based on amplitude and phase components has made it possible to achieve 88% error detection accuracy.</p>