Open-source tools for multi-device vision systems, camera orchestration, and practical capture pipelines.
Vision-Orchestration is a collection of practical, deployable tools for computer vision work — focused on the parts that research repos usually skip: capture, sync, recording, and pipeline infrastructure.
We care about things that actually run on real hardware.
Multi-phone DroidCam controller. Preview, record, and snapshot from up to 10 Android phones simultaneously. Self-healing streams, per-camera write queues, inline session labeling.
Status: Active · Language: Python · Tags: opencv, droidcam, camera-control, recording
Vision-based foot gesture recognition system for industrial environments. Custom multi-view dataset, EfficientNetV2 + R(2+1)D pipeline, 8-class gesture classification.
Status: In development · Language: Python · Tags: deep-learning, gesture-recognition, pytorch
- Multi-camera sync and calibration utilities
- Dataset capture and annotation helpers
- Lightweight inference wrappers for edge deployment
Multi-device camera control → orchestrate phones, webcams, IP cameras
Vision pipelines → from raw capture to trained model
Capture & recording tooling → structured, labeled, production-quality data
Lightweight deployment → CPU/consumer-GPU friendly inference
Good vision tools should be simple to set up, robust in the field, and free.
We build things that work without cloud dependencies, expensive rigs, or enterprise licenses.
- Use a tool → open an issue if something breaks
- Have an idea → open a discussion
- Want to contribute → PRs are welcome on any active repo
Maintained by @luuucciiffeerr