Detection of differentiation in general and adipogenesis specifically is conventionally practised by taking only the few cells into account which are visible in the field of view provided by optical microscopy using high-resolution objectives. Other methods of quantification of adipogenic differentiation such as real time PCR, measurement of glycerophosphate dehydrogenase activity or adipogenesis assays only provide integral information lacking spatial resolution and information on the fraction of differentiated cells. Here we used high-resolution scanning and automated image processing to automatically analyze and quantify cell numbers in the range of 20,000. For optimisation of the approach, human gland stem cells (GSC) were differentiated to the adipogenic phenotype comprising inclusion of lipid vesicles. Oil red O and 4',6'-diamidino-2-phenylindole (DAPI) staining made it possible to derive the number of differentiated cells in relation to the total number of cells. For evaluation of the image processing software we verified our results using adipogenesis assay and phase contrast based cell counting. We developed a method of determining differentiation efficiencies covering the range from 10% down to 100ppm with the same image processing and an identical set of parameters, matching the results of the adipogenesis assay. Our approach is based on a statistically significant number of cells and shows high sensitivity taking into account the heterogeneous differentiation pattern of adipogenesis in GSC and other stem cells.