skills.rest
SharpAI
SharpAIOFFICIAL
Explore all skills by this author →
GitHub

yolo-detection-2026-openvino

Detects objects in camera frames using YOLO models on Intel OpenVINO devices via Docker.

3.0k|478|Updated Mar 5, 2019
Search Tags:#docker#object detection#surveillance#yolo#intel gpu#openvino#ncs2
One-click install
npx skills add https://github.com/SharpAI/DeepCamera --skill yolo-detection-2026-openvino
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: yolo-detection-2026-openvino
Source: https://github.com/SharpAI/DeepCamera/tree/main/skills/detection/yolo-detection-2026-openvino
Command: npx skills add https://github.com/SharpAI/DeepCamera --skill yolo-detection-2026-openvino

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires ultralytics, numpy, Pillow, openvino, and includes scripts (resource) components.

What problem does it solve?

Running real-time object detection on camera streams requires hardware-specific model optimization and runtime setup, which is difficult to configure consistently across Linux, macOS, and Windows machines with different Intel accelerators.

Core Features & Use Cases

  • Cross-Platform Docker Runtime: Builds a local OpenVINO container that runs detection on Intel NCS2 USB sticks, Intel iGPU/Arc GPUs, or any x86_64 CPU with automatic device selection.
  • Configurable Detection Pipeline: Adjust confidence thresholds, target COCO classes, input resolution, frame rate, and model precision (FP16/INT8/FP32) through parameters.
  • JSONL Streaming Protocol: Exchanges frame events and detection results with the host over stdin/stdout, including performance statistics with latency percentiles.
  • Use Case: Connect a home security camera and detect people, cars, and animals at 15 FPS on an Intel NCS2 stick, receiving structured bounding-box events for each frame.

Quick Start

Run the deploy script to build the Docker image and start detecting objects in your camera frames with default settings.

Frequently Asked Questions about yolo-detection-2026-openvino

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run YOLO object detection with OpenVINO?▼

Run the deploy.sh script, which builds a Docker image containing the OpenVINO runtime and detection script. The skill auto-downloads yolo26n.pt and exports it to OpenVINO IR format on first run, then processes frames sent over stdin.

What hardware does OpenVINO object detection support?▼

It supports Intel NCS2 USB sticks (MYRIAD), Intel integrated GPUs, Intel Arc discrete GPUs, and any x86_64 CPU. The AUTO device setting lets OpenVINO pick the fastest available option, with CPU as fallback.

Can I use OpenVINO detection on macOS or Windows?▼

Yes, the skill runs inside Docker Desktop 4.35+ on macOS and Windows. NCS2 passthrough uses Docker Desktop USB/IP on macOS or usbipd-win with WSL2 on Windows, and CPU fallback is always available.

How do I export a YOLO model to OpenVINO IR format?▼

Run scripts/compile_model.py with your model name, image size, and precision, for example FP16 for GPU or NCS2. Unlike Edge TPU compilation, OpenVINO export works on any platform without special hardware.

Why is my OpenVINO detection running in CPU fallback mode?▼

CPU fallback occurs when no GPU or NCS2 device is detected during deployment. Check that /dev/dri and /dev/bus/usb are passed to the container on Linux, and verify device availability with scripts/device_probe.py.

What is the difference between FP16, INT8, and FP32 precision?▼

FP16 is fastest on Intel GPU and NCS2 devices, INT8 is fastest on CPU, and FP32 offers the highest accuracy at lower speed. The default FP16 setting targets accelerator hardware.

Related Skills

Explore More Software Engineering Skills→
Yogi-076Yogi-076

camsnap

Capture frames and short clips from RTSP/ONVIF IP cameras.

Community
Intermediate
joeleung0122joeleung0122

camsnap

Automate frame and clip capture from RTSP/ONVIF cameras using ffmpeg.

Community
Intermediate
SharpAISharpAI
3.0k

HomeSafe-Bench

Benchmark Vision-Language Models on indoor home safety hazard detection.

Official
Intermediate
choice5346choice5346

llama-cpp

Run LLM inference on CPUs, Apple Silicon, and non-NVIDIA GPUs with llama.cpp.

Community
Intermediate