Hello, I am writing a project that should check 2 images and tell if the picture is 1 picture 2.

For example, a picture from meme 1 is a picture 2. Tell me: is it possible to implement this without a neural network, or what network will you have to use?

  • Neural networks are more designed for specific tasks, for example, recognizing the character s or a cat in an image. Still wondering exactly how you want to compare images, pixel by pixel, or, for example, primitives? Will it be specific images or everything? - Jagailo
  • I am interested in the way in which the program can determine whether the image is a clone even if it is cropped or slightly modified (brightness is added, slight distortions when optimizing for the Internet), the method of bundles of neural networks came to mind, but realizing that they are needed, to define an object, I asked a question about the implementation without neural networks. 1 image is target, and the second is either left or target but cropped - Walker
  • So there are so many algorithms for comparing images, there is info on the habr, I'm not talking about English-language topics. Neural networks will also have to tinker, take a look in the direction of Microsoft Cognitive Toolkit . - Jagailo
  • That is, the first image has a resolution of 400x800, and the second one, for example, 400x720, but they are one and the same, so that the program compares them to produce at least that image 2 differs from the first one in x from y - Walker

0