The invention discloses a mechanical response characteristic testing system and method for a cemented filling body. The system comprises a cemented filling body three-axis compressive strength testingdevice, a computer and an SEM scanning electron microscope. The method comprises the following steps: I, preparing an SEM sample; II, forming an SEM scanning image; III, carrying out Gaussian filterprocessing on the SEM scanning image; IV, obtaining a plurality of cemented filling body clustering images; V, determining a cemented filling body microcosmic porous image to obtain a cemented fillingbody microcosmic binary image; VI, obtaining a test sample image; VII, normalizing the test sample image; and VIII, inputting the normalized test sample in a pre-established Tensorflow deep learningmechanical response prediction network to obtain a three-axis mechanical response prediction result. According to the method, steps are simple, the prediction efficiency is high and functions are comprehensive, and the method has important meaning on researching strength and stability of the cemented filling body.