Video Search GlossaryKey Terms Explained
Clear, self-contained definitions for the concepts behind AI video search — so you know exactly what each term means and how it works.
AI video search
AI video search is the use of machine-learning models to look inside a video file and locate specific content, rather than searching only its title, tags, or transcript. The AI analyzes the actual frames and audio so you can find a person, action, object, scene, or spoken phrase and jump straight to the moment it appears.
Video scene search
Video scene search is the ability to find a particular scene inside a longer video by describing what happens in it. Instead of scrubbing through the timeline, you describe the setting or event — for example "the sunset over the beach" — and the tool returns the segments that match, each with its start time.
Moment finder
A moment finder is a tool that pinpoints the exact instant something occurs in a video from a plain-language description. It returns a precise timestamp for each match so you can navigate directly to that moment instead of watching the whole clip.
Natural-language video search
Natural-language video search lets you search a video using everyday sentences instead of keywords or filters. You type a description of what you want to find the way you would say it aloud, and the system interprets your intent to return the matching moments.
Timestamped scene match
A timestamped scene match is a search result that pairs a matching moment with the exact time it occurs in the video. Each result includes the point in the timeline where the content appears, letting you click straight to it rather than searching manually.
On-screen text search (OCR in video)
On-screen text search uses optical character recognition (OCR) to read text that appears within video frames — such as signs, slides, captions, or license plates — and makes it searchable. This lets you find every moment where a specific word or phrase is visible on screen, even when it is never spoken.
Video content search
Video content search means searching the actual visual and audio content of a video rather than its surrounding metadata. It indexes what is seen and heard inside the file so queries match the footage itself, not just the file name or description.
Semantic video search
Semantic video search matches the meaning of your query rather than exact words. It understands concepts and synonyms, so a search for "a person cooking" can surface a clip of someone frying eggs even if no label ever uses those exact terms.
Keyframe
A keyframe is a single representative frame extracted from a video, typically at points where the scene changes. Search and analysis tools sample keyframes to understand a video efficiently without processing every frame, which speeds up indexing and matching.
Reverse video search
Reverse video search is the process of taking an existing video or clip and finding its original source or other places it appears online. It answers "where did this footage come from?" — a different task from searching inside a video for a moment. SearchByVideo does not perform reverse video search; it searches within a video you provide to find specific moments.
Deep search / video indexing
Deep search, or video indexing, is the background step that analyzes an entire video and builds a searchable index of its people, actions, objects, scenes, and text. Once a video is indexed, later searches return results instantly because the heavy analysis has already been done.
Clip export
Clip export is the ability to save a matched moment as its own short video file. After finding a moment, you can trim to that segment and download it, making it easy to share a highlight or reuse a specific portion without the full video.
Try AI video search for yourself
Upload a video, describe the moment you need, and jump straight to the timestamp.
Search a video free