Frame-Rate Up-Conversion (FRUC) is a frame-based video manipulation where synthesized frames are periodic inserted into two consecutive frames to increase the frame-rate of videos. Thus, it can be employed by counterfeiters for falsifying high frame rate videos or splicing videos with different frame-rates. Automatically identifying the adopted FRUC techniques from a serial of faked frame-rate videos while achieving a stable recognition rate is a challenging task in video forensics. We propose a new forensic framework, named Spatial-Temporal Edge-Occlusion Identification of FRUC (STEO-IoF), by considering the capabilities of both clues: spatial-temporal edge and occlusion maps, to guide a forensic model using prior knowledge transferred from the existing steganalysis convolutional neural network (CNN) model for the recognition of adopted FRUC techniques. Three open FRUC softwares and four representative FRUC techniques have been evaluated, and experimental results clearly show that the efficacy of the proposed approach.