Refraction testing is essential in ophthalmology clinics and typically involves using a refractor or retinoscopy while under cycloplegia. Retinal Fundus Photos (RFPs) offer a plethora of knowledge about the human eye and may offer a new, more practical and objective method. Here, our goal was to create and test a Fusion Model-Based Intelligent Retinoscopy System (FMIRS) that would allow us to compare the cycloplegic refraction to ocular refraction using RFPs. The FMIRS was built, and the effectiveness of the sphere and cylinder regression models was assessed. The classification model of the cylinder axis was assessed using the accuracy, sensitivity, specificity, area under the receiver operating characteristic curve, and F1-score metrics. The ocular refraction was successfully and precisely identified by the FMIRS in the sphere, cylinder, and axis, and it demonstrated good agreement with the cycloplegic refraction. The RFPs can offer detailed fundus information as well as information on the eye's refraction state, underscoring their potential clinical value.