What tools can be used to compare animals in pictures?

Question

Python

Sorry to be stupid, but I'm having an annoying idea. Parse various sources where people upload lost animals and sources where they post foundlings, compare them and, if there are matches, send them to the owner. I can parse, I’ll write a simple website, I can also work with the database, but neurons ... I can’t imagine how I can compare animals in 2 photographs, which can be from different angles, with different shadows, and so on. Tell me some tools / libraries / read something for the materiel, or at least in what direction to move. Thanks in advance

Answer 1

The best image classifiers can sort of identify the breed, but no more.
However, the ability to narrow the range for comparison is already something ...
In general, read the image classifiers topic. Maybe there are some ready.

Answer 2

pillow is a good photo comparison library. And fluffy faces, try to do on her. But you have to suffer a little to implement scanning with face comparison

Answer 3

On the fingers, it looks like this:
Neuronka "compresses" a picture of a vector of numbers. During training, it is forced to give vectors close to each other for pictures of the same objects. Those. conditionally, if we take a two-dimensional vector (plane - XY), then the upper right corner will be occupied by white dogs, and black ones in the lower, and something in between - mixed colors.
As a result, it remains only to look for vectors that are close to each other.
In reality, multidimensional vectors with hundreds of dimensions are used. Therefore, there are corners for smaller characteristics - dogs with black ears and a white nose, a striped back and a gray paw.
The problem is usually only in the training material.

Answer 4

In the current realities, this is not possible. The maximum that can be achieved is to guess the breed, but a powerful dataset is needed.

Answer 5

I implemented something similar on Tensorflow + connected it to a telegram bot to recognize animals, fish, flowers, mushrooms, etc. This is a whole conveyor for the formation of a dataset and daily retraining of the model. We can try to implement your task. Here are examples of tg bots:
t.me/FishDetectBot
t.me/FungiDetectBot