Agent Skills › SharpAI/DeepCamera › yolo-detection-2026-coral-tpu-macos

yolo-detection-2026-coral-tpu-macos

GitHub

利用Google Coral Edge TPU在macOS/Linux本地硬件进行实时YOLO目标检测，支持80类COCO物体识别，通过USB连接TPU加速器实现低延迟推理。

skills/detection/yolo-detection-2026-coral-tpu-macos/SKILL.md SharpAI/DeepCamera

Trigger Scenarios

需要调用摄像头或图像进行实时目标检测请求使用Edge TPU硬件加速进行物体识别

Install

npx skills add SharpAI/DeepCamera --skill yolo-detection-2026-coral-tpu-macos -g -y

More Options

Non-standard path

npx skills add https://github.com/SharpAI/DeepCamera/tree/master/skills/detection/yolo-detection-2026-coral-tpu-macos -g -y

Use without installing

npx skills use SharpAI/DeepCamera@yolo-detection-2026-coral-tpu-macos

指定 Agent (Claude Code)

npx skills add SharpAI/DeepCamera --skill yolo-detection-2026-coral-tpu-macos -a claude-code -g -y

安装 repo 全部 skill

npx skills add SharpAI/DeepCamera --all -g -y

预览 repo 内 skill

npx skills add SharpAI/DeepCamera --list

SKILL.md

Frontmatter

{
    "icon": "assets\/icon.png",
    "name": "yolo-detection-2026-coral-tpu-macos",
    "entry": "scripts\/detect.py",
    "mutex": "detection",
    "deploy": {
        "linux": "deploy.sh",
        "macos": "deploy.sh"
    },
    "runtime": "python",
    "version": "1.0.0",
    "category": "detection",
    "parameters": [
        {
            "name": "auto_start",
            "type": "boolean",
            "group": "Lifecycle",
            "label": "Auto Start",
            "default": false,
            "description": "Start this skill automatically when Aegis launches"
        },
        {
            "max": 1,
            "min": 0.1,
            "name": "confidence",
            "type": "number",
            "group": "Model",
            "label": "Confidence Threshold",
            "default": 0.5,
            "description": "Minimum detection confidence — lower than GPU models due to INT8 quantization"
        },
        {
            "name": "classes",
            "type": "string",
            "group": "Model",
            "label": "Detect Classes",
            "default": "person,car,dog,cat",
            "description": "Comma-separated COCO class names (80 classes available)"
        },
        {
            "name": "fps",
            "type": "select",
            "group": "Performance",
            "label": "Processing FPS",
            "default": 5,
            "options": [
                0.2,
                0.5,
                1,
                3,
                5,
                15
            ],
            "description": "Frames per second — Edge TPU handles 15+ FPS easily"
        },
        {
            "name": "input_size",
            "type": "select",
            "group": "Performance",
            "label": "Input Resolution",
            "default": 320,
            "options": [
                320,
                640
            ],
            "description": "320 fits fully on TPU (~4ms), 640 partially on CPU (~20ms)"
        },
        {
            "name": "tpu_device",
            "type": "select",
            "group": "Performance",
            "label": "TPU Device",
            "default": "auto",
            "options": [
                "auto",
                "0",
                "1",
                "2",
                "3"
            ],
            "description": "Which Edge TPU to use — auto selects first available"
        },
        {
            "name": "clock_speed",
            "type": "select",
            "group": "Performance",
            "label": "TPU Clock Speed",
            "default": "standard",
            "options": [
                "standard",
                "max"
            ],
            "description": "Max is faster but runs hotter — needs active cooling for sustained use"
        }
    ],
    "description": "Google Coral Edge TPU — real-time object detection natively (macOS \/ Linux)",
    "capabilities": {
        "live_detection": {
            "script": "scripts\/detect.py",
            "description": "Real-time object detection on live camera frames via Edge TPU"
        }
    },
    "requirements": {
        "platforms": [
            "linux",
            "macos"
        ]
    }
}

Coral TPU Object Detection

Real-time object detection natively utilizing the Google Coral Edge TPU accelerator on your local hardware. Detects 80 COCO classes (person, car, dog, cat, etc.) with ~4ms inference on 320x320 input.

Requirements

Python 3.9–3.13

How It Works

┌─────────────────────────────────────────────────────┐
│ Host (Aegis-AI)                                     │
│   frame.jpg → /tmp/aegis_detection/                 │
│   stdin  ──→ ┌──────────────────────────────┐       │
│              │ Native Python Environment     │       │
│              │   detect.py                   │       │
│              │   ├─ loads _edgetpu.tflite     │       │
│              │   ├─ reads frame from disk     │       │
│              │   └─ runs inference on TPU    │       │
│   stdout ←── │   → JSONL detections          │       │
│              └──────────────────────────────┘       │
│   USB ──→ Native System USB / edgetpu drivers       │
└─────────────────────────────────────────────────────┘

Aegis writes camera frame JPEG to shared /tmp/aegis_detection/ workspace
Sends frame event via stdin JSONL to the local Python instance
detect.py invokes PyCoral and executes natively on the mapped USB Edge TPU
Returns detections event via stdout JSONL

Platform Setup

Linux

# Uses the official apt-get google-coral packages natively
./deploy.sh

macOS

# Downloads and installs the libedgetpu OS payload framework inline
./deploy.sh

Important Deployment Notice: The updated deploy.sh script will natively halt execution and prompt you securely for your OS sudo password to securely register the USB drivers (libedgetpu) system-wide. If you refuse the prompt, it gracefully outputs the exact terminal instructions for you to configure it manually.

Performance

Input Size	Inference	On-chip	Notes
320x320	~4ms	100%	Fully on TPU, best for real-time
640x640	~20ms	Partial	Some layers on CPU (model segmented)

Cooling: The USB Accelerator aluminum case acts as a heatsink. If too hot to touch during continuous inference, it will thermal-throttle. Consider active cooling or clock_speed: standard.

Protocol

Same JSONL as yolo-detection-2026:

Skill → Aegis (stdout)

{"event": "ready", "model": "yolo26n_edgetpu", "device": "coral", "format": "edgetpu_tflite", "tpu_count": 1, "classes": 80}
{"event": "detections", "frame_id": 42, "camera_id": "front_door", "objects": [{"class": "person", "confidence": 0.85, "bbox": [100, 50, 300, 400]}]}
{"event": "perf_stats", "total_frames": 50, "timings_ms": {"inference": {"avg": 4.1, "p50": 3.9, "p95": 5.2}}}