Automatically blur or block sensitive video content while watching in VLC.
Set up best-effort parental controls for shared-screen video playback.
Filter video content locally without sending any frames to a cloud service.
Experiment with ONNX Runtime and GPU-accelerated computer vision inside a media player.
| asayed18/icop | amichail-1/orbination-whisper-ai | andrewrk/libogg | |
|---|---|---|---|
| Stars | 3 | 3 | 3 |
| Language | C | C | C |
| Last pushed | — | — | 2026-03-31 |
| Maintenance | — | — | Maintained |
| Setup difficulty | hard | easy | easy |
| Complexity | 4/5 | 4/5 | 2/5 |
| Audience | general | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Building from source needs CMake, a C/C++17 compiler, VLC development files, and separately downloaded model and runtime files.
icop is a plugin for the VLC media player that adds automatic content filtering to video playback. It uses an AI model running entirely on your own computer to look at each video frame before it is shown, and if the frame is classified as sensitive, the plugin can black it out, blur it, or show a warning instead of the original image. Nothing is uploaded anywhere: all the analysis happens locally through a tool called ONNX Runtime. The plugin works by having VLC hold onto a decoded frame for a moment before showing it. During that pause, icop resizes and converts the frame so the AI model can read it, gets a classification back, and then decides whether to release the real frame or a masked version. The README describes this as a safety rule the whole project is built around: a frame that needs checking should never be shown before the check finishes. The plugin can also mute audio while a blocked frame is on screen, and it supports different block styles and multiple model profiles. icop currently works on Windows and Linux, with an experimental version for macOS. It can use graphics card acceleration where available, falling back automatically between different hardware options and finally to the regular processor if needed. The README is upfront that this is a best-effort filter: the underlying detection model can miss things or flag things incorrectly, so the authors say to test it yourself before relying on it for anything like child safety or accessibility. For people who just want to use it, the README suggests handing the included INSTALL.md file to an AI assistant, which can then handle downloading and setting it up. For developers who want to build it from source, the project uses CMake and a C and C++ compiler, along with the development files for VLC itself, and provides make commands to build, test, and package the plugin. Some parts of the process, like the required model files and prebuilt VLC copies, are downloaded separately rather than stored in the repository. icop is released under the GPL-2.0-or-later license, a copyleft license which generally requires that modified versions of the code also be shared under the same license terms. The full README is longer than what was shown.
A privacy-first VLC plugin that uses a local AI model to detect and block sensitive video frames before they are shown, without uploading anything.
Mainly C. The stack also includes C, C++, ONNX Runtime.
Modified versions of this code must also be released under the same GPL license and made open source.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly general.
This repo across BitVibe Labs
double-check against the repo, no cap.