Abstract

Movie narrating can enable the capture of not only the subject matter but also the emotive essence of the subject. For example, “having a good time” or “feeling exhausted”, which relates to the narrative of “holding a birthday party”. Movie narrating converts the human-like expressivity in movie shots (i.e., sets of film frames) into a cohesive and coherent narrative (i.e., sentences that tell a logical story). Besides being essential for a humanlike understanding of images, movie narrating can realize numerous new applications such as the automatic emotive narrating of photograph albums on social media, automatic logical summary of trips, or diary generation of events. Moreover, as an interdisciplinary field spanning computer vision, natural language processing, and philosophy, movie narrating can potentially elevate artificial intelligence from basic understanding toward human-like understanding.

Download paper here