Python, TensorFlow project

Project goals:

1- build a system that can can detect a specific object.

2- I should be able to use the same system to detect any other object by training it with another images.

3- This code going to be tested on Raspberry pi.

4- The system should be fast because it is going to be running live on the raspberry pi camera.

5- Include all references and fully explain the code in details.

6- This codes should be done in Python 3.

7- system should detect the object in any direction and perform the needed calculation to take the image.

*********************************************************************************

********* After Detecting *****Taking the image*********(This is the main part of the project)

******** This part of the project should work with any other object trained by the first part***

**********************************************************************************

1- Making sure the object is within the limit area in the frame as shown in the attached images.( to make sure there is no missing part from the object) - https://drive.google.com/drive/folders/1UdVkVBdDKkyEZzZ1h7TgegZt1aCfwObI?usp=sharing

2- Creating algorithm to decide whether you need to get closer to the object or not (if object is too small in the frame).(output message should be printed) and system should start from beginning)

3- Creating algorithm to decide whether you need to go farther from to the object if it is too big.(output message should be printed) and system should start from beginning)

4- Taking image when object is within a good size into the live frame and save it in memory which can be adjusted through the codes.

I have built a model that we are going to train the first part with, below you can see two sample images of that object and I will provide more for training.  When testing the model, Raspberry pi camera is going to be used, so images will be in different quality.

For the second part:

Algorithm should take care of it. What I am trying to do is. Take a top view of an object. I am trying to see when the object is more into the center area of the frame and if it is a good size and then takes its image.