Toward Interactively Balancing the Screen Time of Actors Based on Observable Phenotypic Traits in Live Telecast

(Work in progress) Paper (Draft)

Naimul Hoque, Nazmus Saquib, Syed Masum Billah, Klaus Mueller

Several prominent studies have shown that the imbalance in the on-screen exposure of observable phenotypic traits like gender and skin-tone in streaming visual media, such as movies, TV shows, and live telecasts can reinforce gender and racial stereotypes in society. To mitigate this problem, researchers and human rights organizations alike have long been calling for more awareness of these issues in the production of visual media. While awareness among media producers is beginning to grow, there are two key technical limitations in current commercial production software that prevent progress toward these goals. First, most commercial production software lack mechanisms that allow a quantification of the screen time of actors based on their observable phenotypic traits, such as gender and skin-tone in the produced video. Second, there are no visual awareness tools that allow producers to balance the exposure of these phenotypic traits during the production of live telecasts. In this paper, we propose Screen-balancer, an interactive visual tool to overcome these limitations. The design of Screen-balancer is informed by a field study conducted in a professional production studio. Screen-balancer analyzes the facial features of the actors to determine phenotypic traits using off-the-shelf facial detection packages, and then facilitates real-time visual feedback to assist media producers in balancing the screen time of these actors in a live telecast. To demonstrate the effectiveness of our approach, we conducted a user study with 20 participants and asked them to compose live telecasts from a set of video streams simulating different camera angles, and featuring a number of male and female actors with different skin-tones. The study revealed that the participants were able to reduce the difference between male and female screen time by 43% and lighter and darker screen time by 44%, thus showing the promise and potential of using such a tool in commercial production systems.