ContCap: A scalable framework for continual image captioning

Giang Nguyen, Tae Joon Jun, Trung Tran +2

While advanced image captioning systems are increasingly describing images coherently and exactly, recent progress in continual learning allows deep learning models to avoid catast…