Automatic Immersion of Brands in Videos

This demonstration paper presents a partially automatic, human-in-the-loop application for immersing digital brand graphics inside digital videos in an unobtrusive manner. This method automatically identifies regions inside digital videos that are able to host such graphics without obtruding the viewer experience, while discarding regions, including those representing human faces and body parts, that are deemed unable to host such graphics without obtruding the viewer experience. After a human chooses or customizes one of the positively identified regions, and chooses a digital brand graphic to immerse inside it, this method automatically immerses the graphic inside the digital video, taking steps to emulate the occlusion, shadows and light changes, texture, and pose that affected the original region inside the digital video in order to make the immersion feel more realistic and seamless.