Tensor Flow VS Auto ML VS TensorFlow Object Detection API

Question

The question is very complicated for me and I will be grateful for any information.

I need to train a model, but there seem to be so many tools that I am getting confused.

I need to train a model that I will convert to “tensor flow lite” in the future for use in an android app on my phone. I am planning to use ML-KIT vision. My task is as follows - I need to detect an object, then the object should be trimmed on the edges, and already on the object itself to find and recognize the text. Everything is more or less clear with the second part, as there is a ready ML KIT tool from google. With the first part I don't know how to be, because I need a custom model.

1)I found AutoML Vision Edge, but it is outdated and doesn't work at all.

2)I found the AUTO ML in Vertex AI tool, and from what I understand, you can only use it through google cloude, which is accessed for a fee. I don't have that kind of facilities. So I am looking for another way out, for creating a model.

3)I also found TensorFlow Object Detection API, but even there in the repository they write that they will no longer support it.

4)I understand that the only way to create a model through tensor flow directly?

Please give me some hints on which tool is better to use, maybe some other than the above. The thing is that I was originally looking for the simplest tool possible, without using customizations, because I don't know how to customize, and it's still very difficult for me. But if you give me a link to some detailed guide, I will be very grateful.