Various embodiments set forth systems and techniques for evaluating media content items. The techniques include receiving visual feedback associated with one or more audience members viewing a first media content item; analyzing the visual feedback to generate one or more emotion signals based on the visual feedback; and generating a set of features associated with the one or more audience members viewing the first media content item based on the one or more emotion signals.
BACKGROUND Field of the Various Embodiments (1) Embodiments of the present disclosure relate generally to computer science and video analysis and, more specifically, to a behavior-based computer vision model for content selection. Description of the Related Art (2) Media providers, such as television stations, television networks, and streaming services, typically select shows, movies, and other content that will be offered by the media provider through a multi-phase process that involves significant amounts of research and surveys, such as focus groups, dial testing, surveys, and the like. A final decision is made at an executive level after considering the reports resulting from this multi-phase process. (3) One approach for evaluating media content is to use machine learning models to generate predictions of audience reception of the media content. The machine learning models receive media content as input and generate output that indicates a predicted audience appeal of the media content. However, machine learning models typically only evaluate features and characteristics of the media content to generate a prediction. The machine learning models do not account for other factors that relate to audience appeal, such as audience responses to the media content and audience preference. Accordingly, the predictions generated by the machine learning models are not effective or accurate reflections of audience appeal. Therefore, such predictions cannot be relied upon for making
1. A computer-implemented method for evaluating media content items, the method comprising: receiving visual feedback associated with one or more audience members viewing or listening to a media content item; analyzing the visual feedback to generate one or more emotion signals based on the visual feedback, wherein each of the one or more emotion signals quantifies a strength of a specific emotion expressed by a corresponding audience member while viewing or listening to the media content item; generating a set of features associated with the one or more audience members based on the one or more emotion signals; and generating, via a trained machine learning model, a success metric associated with the media content item based on the set of features, wherein the success metric indicates a predicted success of the media content item with one or more other audience members. ||
12. One or more non-transitory computer-readable medium including instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of: receiving visual feedback associated with one or more audience members viewing or listening to a media content item; analyzing the visual feedback to generate one or more emotion signals based on the visual feedback, wherein each of the one or more emotion signals quantifies a strength of a specific emotion expressed by a corresponding audience member while viewing or listening to the media content item; generating a set of features associated with the one or more audience members based on the one or more emotion signals; and generating, via a trained machine learning model, a success metric associated with the media content item based on the set of features, wherein the success metric indicates a predicted success of the media content item with one or more other audience members. ||
20. A system comprising: one or more memories storing instructions; and one or more processors that are coupled to the one or more memories and, when executing the instructions, are configured to: receive visual feedback associated with one or more audience members viewing or listening to a media content item, analyze the visual feedback to generate one or more emotion signals based on the visual feedback, wherein each of the one or more emotion signals quantifies a strength of a specific emotion expressed by a corresponding audience member while viewing or listening to the media content item, generate a set of features associated with the one or more audience members based on the one or more emotion signals, and generate, via a trained machine learning model, a success metric associated with the media content item based on the set of features, wherein the success metric indicates a predicted success of the media content item with one or more other audience members.