One embodiment facilitates identification of re-photographed images. During operation, the system obtains a sequence of video frames of a target object. The system selects a frame with an acceptable level of quality. The system obtains, from the selected frame, a first image and a second image associated with the target object, wherein at least one of a zoom ratio property and a size property is different between the first image and the second image. The system inputs the first image and the second image to at least a first neural network to obtain scores for the first image and the second image, wherein a respective score indicates a probability that the corresponding image is re-photographed, wherein a re-photographed image is obtained by photographing or recording an image of the target object. The system indicates the selected frame as re-photographed based on the obtained probabilities.