For example: The admin would be able to create a "face annotation" project, and decide that in each image the users need to use a polygon tool to mark the entire face, place points on eyes and mouth, and select from a dropdown menu if this is male or female. The code needs to be flexible so that such schemes can be easily defined by the admin in a web-interface or using simple xml (or similar). The admin can also define how many times each image should be annotated. Annotation templates should allow some sort of logic, for example if selecting "male", the admin can decide that the user have to answer another question (e.g. has beard? yes/no) or perform another annotation task (e.g. tag beard with polygon)
Once the project is created and the annotation scheme defined, the admin can upload images and associate users with the project. When a user logs in, they are served with one of the images which was not yet annotated, and they can click "face" to annotate the face with a polygon, click "left eye" to annotate the left eye with a point, answer questions etc.
viewing/grabbing the results should be easy.
I will add you to our git repo and start the contract with a phone call explaining the scope of the project and the features required in more detail.
Hopefully this short project can evolve into a longer gig - there's way more stuff we need to implement!