Lightweight, short video clip advertisements can be embedded at various points within a video content stream and drive brand recall for the advertiser. However, such ads can interrupt the user’s viewing experience and can potentially create a negative impact on the advertised brand. This disclosure describes techniques to place short advertisements within a video content stream at positions where viewing experience is least likely to be negatively affected by such advertisement insertion. The timing of an ad in a video, as originally inserted, is modified by applying a correction based on the audiovisual information from the video stream. Optimal positions to insert advertisements are determined by a trained machine learning model that generates an event score at different points in a video. The points with low event scores and that are within a threshold of the original ad position can be selected for insertion of the advertisement.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Shin, D, "Minimally Disruptive In-stream Advertisement Placement Within Videos", Technical Disclosure Commons, (January 07, 2022)