Surgical Instrument Sample App#

GitHub Readme

Note

This application is for reference and evaluation purposes only. It is not intended for direct use in clinical or diagnostic environments and is not validated for such a purpose.

The app demonstrates how Intel hardware acceleration (CPU / Intel iGPU / Intel NPU) may be applied through OpenVINO to AI-based real-time polyp detection in a video of an endoscopic procedure. It is packaged as a Docker Compose workload with a decoupled capture / inference / display architecture, so the displayed video stays smooth regardless of inference speed.

Supported inputs:

  • Basler industrial camera (via pypylon)

  • USB / V4L2 webcam

  • Video file (for demos and benchmarking)

Display path:

  • OpenGL vsync presenter with cv2 and headless fallback

  • Optional fullscreen direct-scanout for low latency

  • Optional camera trigger modes (software, vsync) for low photon-to-pixel latency

How to start:

  • See Model Preparation if you do not already have a trained OpenVINO IR under models/yolo11n_polyp/best_openvino_model/. Skip this page when pulling a prebuilt model from the registry.

  • See Get Started for a step-by-step deployment guide.

  • See Runtime Configuration for every knob and CLI option exposed by the app.

  • See Troubleshooting for common issues.

  • See Release Notes for version history.