Thermography can be used for nondestructive detection of surface and subsurface defects in objects. Generally, thermal images are recorded with an infrared imager. However, infrared imager cannot work accurately when the inspected surface has a high reflectivity. This paper proposes a fluorescence thermography method in visible light range based on a smartphone, with a principle of fluorescence integral intensity ratio temperature sensing. Considering that the R, G, and B components of each pixel in a fluorescence image are the integral intensity of partial fluorescence spectrum, the intensity ratio of any two of components R, G and B can also reflect temperature according to the spectroscopic principle. Therefore, temperature-dependent fluorescence images can be calibrated for thermal imaging. A YAG:Ce phosphor-based composite membrane was used as the temperature probe, and a smartphone was used to take fluorescence photos and efficiently convert the photos into thermal images through a self-developed APP. The smartphone-based fluorescence thermal imager was verified by applications of defect inspection. A shielded hot soldering iron and dummy defected samples were successfully inspected with the imaging system. Compared with the technique of infrared thermography, the smartphone-based fluorescence thermography has the advantages of applicability to argenteous metals, low cost, and high spatial resolution.