Facial Recognition computer vision application -- 2

I am working on computer vision project related , Facial Recognition. Once we Train our model and test them , it works as expected. However our requirements are little different. Suppose we need to add new face to existing system then , we need to re-train our model [ old data + new data ] and this is the waste of time and resources. we want to develop in such a way that , once new image are added , our model has to train only newly added images . we are looking for some expert to help on this project .