How to divide a graph into parts using a neural network?

Question

Python

I am trying to split a graph into parts. Graphs are different, but I need to highlight certain patterns.
Tried using TensorFlow Object Detection API . But the COCO class library that is used there is not about graphics, but about real life objects (dogs and cats, cars, etc.). Either you need to write your own class with a chart pattern (I don’t even know how to do it), or use another method.
The graph needs to be cut something like this (to get 6 parts). Red - cut line (object boundaries).

We use a neural network because the graphs can be different, for example, like this. And this chart needs to be cut into 4 parts in order to extract the necessary patterns.

Answer 1

There is an interesting package of libraries and a graphical interface Weka , which contains a huge number of algorithms for working with data, searching for patterns, and so on.
Your task is called Classification. Google based on this, and neural networks in this are only one of the options, and not the most effective.
ps 99% of the work - making a decision on the form of data submission and preparing (converting) to enter the algorithm. Those. you can try to draw a graph on a raster and slip this raster into a neural network (and pay just endless money for training), or you can come up with a function through which you run your data to make it more representative or hide flaws (for example, extrema to infinity), ..