Abstract: A machine learning method includes: processing a first global image and a first local image by a first augmentation pipeline to generate a first augmentation global image and a first augmentation local image; processing the first global image by a second augmentation pipeline to generate a second augmentation global image; processing the first augmentation global image to generate a first global representation vector; processing the second augmentation global image and the first augmentation local image to generate a second global representation vector and a first local representation vector; comparing the first and second global representation vectors to generate a global loss function; comparing the first global representation vector and the first local representation vector to generate a mixed loss function; and adjusting the second encoder according to the global loss function and the mixed loss function.

Abstract: A machine learning method includes: distinguishing foregrounds and backgrounds of a first image to generate a first mask image; cropping the first image to generate second and third images; cropping the first mask image to generate second and third mask images, wherein a position of the second mask image and a position of the third mask image correspond to a position of the second image and a position of the third image, respectively; generating a first feature vector group of the second image and a second feature vector group of the third image by a model; generating a first matrix according to the first and second feature vector groups; generating a second matrix according to the second and third mask images; generating a function according to the first and second matrices; and adjusting the model according to the function.

Abstract: An encryption and decryption system is provided, which includes a transmitting device and a receiving device. The transmitting device is configured to store original images and a correspondence table. The receiving device is connected to the transmitting device for storing the correspondence table, wherein the transmitting device generates an encrypted string. The transmitting device selects a representative morphed image from multiple morphed images according to the encrypted string. The transmitting device transmits the representative morphed image to the receiving device and does not transmit the encrypted string to the receiving device. The receiving device recognizes the first original image serial number and the second original image serial number from the representative morphed image. The receiving device looks up the correspondence table according to the first original image serial number and the second original image serial number to generate the encrypted string.