Objection detection with pre-trained model

I have a problem with my work,someone to help me?I want to build a object detection model with object detection api,I’m using a ssd pre-trained model but I’ve -100 img,as it’s not a simple classifcation I don’t know what I must do,I’ve written the code with a tuto but since 2 days I’m debugging,hardly I find a good tuto
link of notebook