Information processing device

An information processing device comprising:

• one or more processors;
• and
• a memory storing one or more programs, wherein
• when executed by the one or more processors, the one or more programs cause the one or more processors to:
• electrically acquire a first image from a first communication device connected to the information processing device via a network, the first image comprising a first portion indicating a first item which is a first product, a second portion different from the first portion, and a third portion different from the first portion and the second portion, wherein the second portion indicates a second item different from the first item, the third portion indicates a third item different from the first item and the second item,
• apply a first machine learning model to the first image according to a predetermined first change type, which is stored in the memory, to change the second item to a fourth item different from the first item, the second item and the third item, to generate a first changed image in which the second item indicated by the second portion in the first image is changed to the fourth item, and the first portion and the third portion in the first image are not changed,
• apply a second machine learning model to the first image according to a predetermined second change type, which is stored in the memory, to change the third item to a fifth item different from the first item, the second item, the third item and the fourth item, to generate a second changed image in which the third item indicated by the third portion in the first image is changed to the fifth item, and the first portion and the second portion in the first image are not changed,
• electrically provide a plurality of second communication devices, which are connected to the information processing device via the network, with the first changed image having been generated by the first machine learning model and the second changed image having been generated by the second machine learning model,
• collect a number of accesses by the plurality of second communication devices to the first changed image and the second changed image,
• specify, based on the collected number of accesses, a changed image which is either one of the first changed image or the second changed image having a higher number of accesses,
• specify, based on the specified changed image having the higher number of accesses, a changed type which is either one of the first change type or the second change type corresponding to the specified changed image having the higher number of accesses, and
• apply a third machine learning model to a second image including a fourth portion indicating a second product different from the first product, a fifth portion indicating the second item, and a sixth portion indicating the third item, to generate a third changed image in which either one of the second item indicated by the fifth portion or the third item indicated by the sixth portion in the second image is changed according to the specified change type having the higher number of accesses, and the fourth portion and another one of the second item indicated by the fifth portion or the third item indicated by the sixth portion in the second image are not changed,
• wherein no image, in which the another one of the second item indicated by the fifth portion or the third item indicated by the sixth portion in the second image is changed according to another change type which is either one of the first change type or the second change type having a lower number of accesses based on the collected number of accesses, is generated, and
• the first machine learning model, the second machine learning model and the third machine learning model are generative models and generate the first changed image, the second changed image and the third changed image, respectively, using an algorithm of generative adversarial network.