videosegment

A videosegment is a temporally bounded portion of video content, defined by a start time and an end time within a source file. It serves as a basic unit for organizing, indexing, retrieving, and processing video data. A videosegment typically carries metadata such as duration, source identifier, and optional annotations like labels, categories, or confidence scores describing its content.

Segments are used in many workflows, including content authoring and editing, video search and retrieval, and streaming. In indexing, segments are annotated with semantic information to support queries by topic, event, or scene. In playback and streaming, media is delivered in segments corresponding to time intervals, enabling progressive download and adaptive bitrate.

Types of videosegments include shot-based segments, which align with camera cuts; scene-based or event-based segments, which span broader or more semantically meaningful intervals; and arbitrary or user-defined segments for specialized tasks. Segments can be created manually by annotators or automatically by algorithms such as shot boundary detection, scene change detection, or action recognition. Hybrid approaches combine automatic detection with human verification.

A videosegment is distinct from a single frame or from a shot, though it may coincide with them or cross their boundaries. Its granularity is chosen to balance precision and practicality for the task at hand. In many standards and tools, a videosegment is represented by a pair of timecodes (start and end) and may be associated with metadata in a database or metadata track.