Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

VideoHighlighter

github.com

  • Free plan
  • Windows
  • macOS
  • Linux
7.7 Club score Visit github.com

What VideoHighlighter is

VideoHighlighter is a local desktop app that analyzes videos and exports highlight reels or individual clips. It scores scenes using motion, audio, objects, actions, and transcript signals, then shows the reasoning in a timeline and report. Reports also include per-signal point breakdowns, strong moments that missed the cut, and transcript claims that were not measured. Video files, transcripts, detections, and local models stay on disk; analysis, search, and export do not upload footage. The basic pipeline needs no API key, and a GPU is optional because current CPUs can run it. The free edition includes offline analysis, live face detection, VR side-by-side playback, CLIP search, composition rules, model training, and a model hub. Pro adds object and action overlays, point-and-teach categories, similarity search, open-vocabulary detection, and counter or scoreboard detection. Prebuilt downloads are available for Windows and macOS; Linux runs from source with pip and Python. Free software is released under AGPL-3.0, with a separate commercial license offered. Individual Pro costs $49, paid once.

Who it suits

VideoHighlighter suits editors who want to analyze local footage and produce clips while keeping video and analysis on disk. It also supports YouTube and Twitch VOD inputs.

Good to know, and what to watch

  • Analysis, search, and export do not upload footage.
  • Basic pipeline requires no API key.
  • Free edition includes offline analysis and CLIP search.
  • GPU is optional; current CPUs can run the pipeline.
  • Linux runs from source with pip and Python.
  • Pro activation requires an internet connection once.
  • Individual Pro covers one editor on up to five computers.

Studio

—

Free

Free

Individual

$49 one-time

Free plan

Getting started

  1. Open the VideoHighlighter GitHub website
  2. Download the prebuilt Windows or macOS application
  3. Install from source with pip and Python on Linux
  4. Import a local video, YouTube VOD or Twitch VOD
  5. Run analysis and review the timeline or report
  6. Export highlight reels or individual clips

Limits to check

The free plan is provided at no cost and includes the listed offline features. Individual Pro is for one editor and may be used on up to five computers.

The Club's review

VideoHighlighter is a local desktop application for turning footage and supported VODs into scored highlight reels and individual clips. It suits editors, creators and privacy-conscious teams that want to inspect the reasoning behind selections rather than accept an unexplained result. Its free edition is unusually broad, while Pro mainly serves advanced detection and commercial workflows.

Overview

The app evaluates scene, motion, audio, objects, actions and transcript signals, then presents the results on a timeline and in a report. Reports show each signal's point contribution, which signals fired, strong moments that failed to make the cut, and transcript claims that were not measured. That makes it practical for editors who need to audit or refine a reel instead of simply exporting whatever an automated scorer chooses.

Video files, transcripts, detections and local models remain on disk. Analysis, search and export do not upload footage, and the basic pipeline needs no API key. A GPU is optional: current CPUs can run the pipeline, while Intel Arc and other GPUs can accelerate detection. This is a good fit for sensitive footage and offline work, but less suitable when a team expects a shared cloud workspace.

Key features

The technical stack combines YOLOX with OpenVINO, YuNet and SFace face models, Whisper transcription, a composition engine, CLIP search and a PySide6 editor. The free edition includes offline analysis, live face detection, VR side-by-side playback, CLIP search, composition rules, model training and a model hub. That is a substantial toolkit for a no-cost editor, especially for users willing to tune models and composition rules themselves.

Pro adds live object and action overlays, point-and-teach categories, find-more-like-this search, open-vocabulary detection, and counter or scoreboard detection. Those additions matter for sports, events and specialist footage where faces alone are not enough, but they are unnecessary if basic scoring and clip export cover your workflow.

Inputs include local video files, YouTube VODs and Twitch VODs. YouTube support, video uploads, transcript output and chapter generation make the application useful beyond purely local files, while its local-processing model keeps the analysis stage on the computer.

Pricing

Free — free. This AGPL-3.0 edition includes offline analysis and runs on Windows, macOS and Linux. It is the sensible starting point for personal projects, experimentation and editors who do not need the Pro-only detectors or a commercial license.

Individual — $49, paid once. The one-time purchase covers a single editor on up to five computers, adds a commercial license and priority support, and unlocks all Pro capabilities. It is attractive for a solo professional who wants advanced detection without a recurring renewal. The one-editor restriction means it is not a shared-seat solution, even though it can be installed across five machines.

Studio — custom pricing. This tier is for teams and multi-seat deployments, with multiple seats and volume pricing. Organizations should choose it when Individual's single-editor terms do not fit; the published price is not a basis for comparing total team cost.

Pro activation needs an internet connection once. After activation, the license is checked locally and the app does not contact the vendor. Support comes through Discord and email, with priority treatment for Pro customers rather than a ticket portal.

Platforms

Prebuilt downloads are available for Windows and macOS. Linux runs from source with pip and Python, so Linux users should be comfortable with a more involved installation path. The app supports local files plus YouTube and Twitch VOD inputs, and a GPU is optional.

Who it's for

VideoHighlighter is best for individual editors, creators and production teams handling footage that should remain local, particularly when explainable scoring is important. The free plan also suits learners and hobbyists exploring face detection, CLIP search, model training and composition rules. Choose Pro for commercial work, one-editor businesses, or projects that depend on open-vocabulary, action, object, scoreboard or counter detection. Look elsewhere if you need a browser-first collaborative workflow, multiple Individual editors, or a simple cloud summarizer.

Pros and cons

Pros

  • Local by design: footage, transcripts, detections and models stay on disk, and analysis, search and export do not upload video.
  • Explainable selections: signal-level scores, fired signals and missed strong moments help editors judge and adjust automated choices.
  • Generous free edition: offline analysis, face detection, VR playback, CLIP search, composition rules, training and the model hub are included at no cost.
  • Flexible one-time Pro purchase: Individual adds commercial rights and advanced detection without a subscription renewal.

Cons

  • Linux installation is manual: source, pip and Python are required instead of a prebuilt download.
  • Individual Pro is single-editor only: five-computer use does not provide five seats.
  • Support is informal: Discord and email replace a conventional ticket portal.
  • Advanced capabilities cost extra: live object and action overlays, teaching categories and specialist detection are reserved for Pro.

Alternatives

For a wider comparison, see AI Video Summarizers.

Distill is a better choice for iOS users who mainly need one daily YouTube summary and accept a 30-minute, captions-required limit on the Guest plan. Choose sipsip.ai for a web workflow with subscription slots, transcription credits and AI summaries or translation. SonicScribe suits browser users who want three daily transcripts, 30-minute uploads and exports on its free plan.

Summarize Video to Text is preferable when Android, iOS, a browser or an extension matters, with two no-account videos daily or ten monthly credits after Google sign-in. Claras fits extension and web users needing playlist search, though its free plan allows three videos monthly. Mindgrasp is aimed at study materials and questions about uploaded content. Video Highlight offers extension and web access with roughly 7 hours 30 minutes of public-video processing on its free plan. Eightify is a paid-oriented option for browser, mobile and extension users who want concise summaries.

Verdict

Choose VideoHighlighter if you want local, explainable video analysis and a capable free starting point, or if a solo commercial editor needs Pro's specialist detection for a one-time $49 purchase. Look elsewhere for cloud collaboration, simpler installation, multiple editors under one license, or a primarily transcript-and-summary workflow.

The verdict

VideoHighlighter offers local analysis with explanations for its highlight decisions and a useful free edition. Check the Linux installation method and the one-editor Pro license limit before choosing it.

Questions people ask

Which platforms are supported?

VideoHighlighter supports Linux, macOS and Windows. Prebuilt downloads cover Windows and macOS, while Linux runs from source with pip and Python.

Does it upload video footage?

No. Video files, transcripts, detections and local models stay on disk, and analysis, search and export do not upload footage.

What does Pro add?

Pro adds live object and action overlays, point-and-teach categories, find-more-like-this search, open-vocabulary detection and counter or scoreboard detection.

Is an API key required?

The basic pipeline needs no API key.

What the maker publishes

Activation
Pro activation requires an internet connection once, after which the license is checked locally and the app does not contact the vendor. source
API requirement
The basic pipeline needs no API key. source
Detection
The app scores scene, motion, audio, objects, actions and transcript signals, then displays the reasons on a timeline and report. source
Explainability
Reports provide per-signal point breakdowns, fired signals, strong moments that missed the cut and transcript claims that were not measured. source
Free capabilities
The free edition includes offline analysis, live face detection, VR side-by-side playback, CLIP search, composition rules, model training and a model hub. source
Hardware
A GPU is optional; current CPUs run the pipeline, while Intel Arc and other GPUs can accelerate detection. source
Install platforms
Prebuilt downloads are available for Windows and macOS, while Linux runs from source with pip and Python. source
License
The free repository is released under the GNU Affero General Public License v3.0, with a separate commercial license offered by the copyright holders. source
Offline privacy
Video files, transcripts, detections and local models stay on disk; analysis, search and export do not upload footage. source
Pro capabilities
Pro adds live object and action overlays, point-and-teach categories, find-more-like-this search, open-vocabulary detection and counter or scoreboard detection. source
Purpose
VideoHighlighter is a local desktop app that analyzes video and exports highlight reels plus individual clips. source
Support
Support is provided through Discord and email, with priority support for Pro customers and no ticket portal. source
Technical stack
The pipeline uses YOLOX with OpenVINO, YuNet and SFace face models, Whisper transcription, a composition engine, CLIP search and a PySide6 editor. source
Usage limit
An Individual Pro license is for one editor and may be used on up to five computers. source

Checked against the maker's pages on 1 Oct 2026.