A smart video content analysis system for content categorization and highlights generation
Game Video Analyzer (GVA) leverages multi-modal technologies based on signal processing and deep learning to analyze visual, audio, text, and other modal information in game videos. It utilizes deep learning based time series modeling technology to model high-level semantic information to understand video content, detect intense scenes, and automatically generate video highlights.
GVA allows you to customize the detection of key events according to different rules for different games based on game genres. Output video duration can also be customized as needed, which facilitates integration with various systems.
GVA can automatically recognize game content, edit and generate highlights, and add hierarchical tags in batches, significantly reducing your labor costs and improving production efficiency.
GVA is pay-as-you-go and can be run in either CPU or GPU, helping you balance processing speed and costs.
GVA detects start/end times as well as event types, such as kills and weapon selection, enabling game event analysis in deeper detail. In addition, it can determine the scene intensity by evaluating the audio as well as on-screen comments and tag the video accordingly.
GVA can generate a thumbnail for each video highlight to form a set of video covers for you to select manually or loop through. It can also generate animated images in .gif and other formats to function as video covers displayed during mouse over or user interaction.
GVA can automatically generate granular hierarchical tags such as game genres, game names, and event actions. It supports custom tagging as well, making it easier for you to manage game media assets.
GVA can automatically generate highlight reels by selecting prominent scenes such as shootings, narrow escapes and kills from live or recorded gaming streams. These reels can be used as marketing or promotional materials, reducing labor costs coming from human screening and editing.
GVA can automatically generate post-game highlights from esports competitions for timely news coverage and highlights playback. Clips are auto-tagged by intensity levels and placed accordingly on the game video websites, surfacing the most compelling content to the top, thus improving overall browsing experience.
GVA can automatically generate short highlight clips from lengthy game videos for push notifications, reducing the costs of human editing and screening. Video contents can be auto-tagged according to game or game genre. Targeted marketing can be achieved via pushing the tagged content according to users’ game or game genre preferences determined via user behavior analysis.