Image detection method and device based on dual-function discriminator

The invention discloses an image detection method and device based on a dual-function discriminator. The method comprises the following steps: preprocessing a to-be-detected image pair; inputting the pre-processed image pair into a pre-constructed dual-function discriminator network, down-sampling the input image pair into image pairs of different scales by using a 2D average pooling layer through a multi-scale scaling module, and obtaining scaled image pyramid data; extracting deep features from the scaled pyramid data through a feature extraction module; the full convolutional network in the dual-function discriminator is used for carrying out true and false detection on the image pairs, and the multi-classification network is used for carrying out similarity detection on the image pairs. According to the method, the dual-function discriminator network is adopted, so that the method can be suitable for rapid detection of the images, authenticity and similarity of the two images can be efficiently and accura