Build Deep Learning (DL/AI) program and api for object detection (image and video)

In this we require a project to detect 20-30 different landmarks/labels/tags from image and video of mouth (intra oral picture and video) , and use it to train a model which is able to detect those in images and video. The program is to be linked with mobile application so will require an api for that. The accuracy required will be good (enough sample data will be available).
In attached images please see what kind of objects are needed to be detected from images/video.