Detection system + deciding when to take image

Hi everyone, I have this project and would like some help, Please read through: 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 ****************(This is the main part of the project) 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 project) 2- Creating algorithm to decide whether you need to get closer to the object if it 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 farther from to the object if it is too 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. Send me a message if you need any other information about the project, thanks.