← 返回 Skills 市场
13681882136

Kami Image Search

作者 KamiVision · GitHub ↗ · v1.0.1 · MIT-0
cross-platform ⚠ suspicious
34
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install kami-image-search
功能描述
A skill by Kami SmartHome. Capture frames from your camera, describe them with AI, and search your visual history using natural language.
使用说明 (SKILL.md)

\r \r

Kami Image Search\r

\r

Search your camera's visual history with natural language.\r \r Monitor your camera feed, capture frames automatically, and retrieve matching images by simply describing what you're looking for. Powered by the Kamivision cloud API for AI description and embedding generation.\r \r

Features\r

\r

  • 🔍 Natural language image search\r
  • 📸 Automatic frame capture from video streams\r
  • 🧠 AI-powered image description (Kamivision API)\r
  • 📁 Manual image import (JPEG / PNG / BMP / WebP)\r
  • 🔁 Built-in duplicate frame detection\r
  • ⏱ Time-range filtering\r
  • 🏠 Designed for Kami SmartHome ecosystem\r \r

Scenarios\r

\r

  • Doorstep delivery verification\r
  • Home activity review\r
  • Pet or child monitoring playback\r
  • Batch photo indexing and retrieval\r \r

Installation\r

\r

bash setup.sh\r
```\r
\r
Checks for `python3` and `ffmpeg`, creates `.venv/`, and installs `opencv-python-headless`, `numpy`, `requests`, `Pillow`, `faiss-cpu`. Idempotent.\r
\r
## Prerequisites\r
\r
- `python3` and `python3-venv` installed\r
- `ffmpeg` installed (setup.sh will attempt to install it)\r
- `image_config.json` configured with your stream URL and Kamivision API key\r
- RTSP/RTMP camera online and reachable (for capture mode)\r
\r
## Kamivision API Key Setup\r
\r
This skill requires a `KAMIVISION_API_KEY` to access the Kamivision cloud API. The key is NOT included in the skill package — you must provide your own.\r
\r
**Ask the user: Do you already have a `KAMIVISION_API_KEY`?**\r
\r
- **Yes** → Enter the key directly. It will be saved to `image_config.json`.\r
- **No** → Please register at [Kamivision Flow](https://kamiclaw-skill.kamihome.com/) to obtain your API key, then come back and enter it.\r
\r
## Parameter Confirmation\r
\r
Before running, confirm these key settings in `image_config.json`:\r
\r
| Parameter | Default | Description |\r
|-----------|---------|-------------|\r
| `STREAM_URL` | — | RTSP/RTMP/HTTP camera stream URL |\r
| `DEVICE_ID` | `CAM-001` | Camera device identifier |\r
| `KAMIVISION_API_KEY` | — | Your Kamivision API key |\r
| `CAPTURE_INTERVAL` | `10` | Seconds between frame captures |\r
| `SIMILARITY_THRESHOLD` | `0.35` | Search similarity threshold (0.0–1.0) |\r
| `SEARCH_TOP_K` | `5` | Max results per search |\r
| `TIME_ZONE_OFFSET` | `0` | UTC offset in hours for local time display (e.g. `-12` for UTC+12, `8` for UTC+8) |\r
\r
**Ask the user: do any parameters need to be changed?**\r
\r
## Usage\r
\r
### Start Capture\r
\r
```bash\r
.venv/bin/python image_search.py --start-capture\r
```\r
\r
### Stop Capture\r
\r
```bash\r
.venv/bin/python image_search.py --stop-capture\r
```\r
\r
### Check Status\r
\r
```bash\r
.venv/bin/python image_search.py --status\r
```\r
\r
### Import Images\r
\r
```bash\r
# Single image\r
.venv/bin/python image_search.py --import /path/to/photo.jpg --json\r
\r
# Entire directory (recursive)\r
.venv/bin/python image_search.py --import /path/to/photos/ --json\r
```\r
\r
### Search\r
\r
```bash\r
.venv/bin/python image_search.py --search "keys on the table" --json\r
```\r
\r
### Search with Time Range\r
\r
```bash\r
.venv/bin/python image_search.py \\r
  --search "person in blue jacket" \\r
  --time-start 1754538000 --time-end 1754541600 \\r
  --json\r
```\r
\r
## Output (stdout JSON)\r
\r
```json\r
{\r
  "status": "ok",\r
  "query": "keys on the table",\r
  "count": 1,\r
  "results": [\r
    {\r
      "image_name": "CAM-001_1754538507.jpg",\r
      "image_path": "/opt/image_data/CAM-001/20250815/CAM-001_1754538507.jpg",\r
      "description": "A set of keys and a wallet on a table",\r
      "timestamp": 1754538507,\r
      "time": "2025-08-15 10:00:07 AM",\r
      "score": 0.7823\r
    }\r
  ]\r
}\r
```\r
\r
## Exit Codes\r
\r
| Code | Meaning |\r
|------|---------|\r
| `0` | Success |\r
| `1` | Error (config issue, stream failure, API error, runtime exception) |\r
\r
## Troubleshooting\r
\r
- `bash: .venv/bin/python: No such file or directory` → Run `bash setup.sh`\r
- `OpenCV cannot open stream` → Check camera is online and `STREAM_URL` is correct\r
- `Kamivision API error` → Verify `KAMIVISION_API_KEY` and network connectivity\r
- `Unsupported file format` → Only JPEG, PNG, BMP, WebP are supported\r
- `No search results` → Ensure images have been captured/imported; try lowering `SIMILARITY_THRESHOLD`\r
- `FAISS index load failed` → Index may be corrupted; system rebuilds automatically, re-import data if needed\r
安全使用建议
Review and edit `image_config.json` before use. Replace the bundled `STREAM_URL`, verify the cloud API endpoint, understand that images may be uploaded for AI processing, set retention/storage controls, and only start background capture when you intend ongoing monitoring.
能力标签
cryptorequires-walletrequires-sensitive-credentials
能力评估
Purpose & Capability
The camera capture, AI description, local indexing, and natural-language search features match the stated purpose, but they inherently handle sensitive home-camera imagery.
Instruction Scope
The included config contains a concrete RTSP URL with an embedded admin password rather than a blank placeholder, while the skill is documented as requiring the user to configure their own stream.
Install Mechanism
Setup is user-directed but uses a shell script to install ffmpeg and unpinned Python dependencies into a virtual environment.
Credentials
Camera access, recursive image import, cloud API calls, and local image storage are proportionate to visual-history search, but users should scope them carefully.
Persistence & Privilege
The skill supports background frame capture and persistent local indexes, with documented status and stop commands and a retention setting.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install kami-image-search
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /kami-image-search 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.1
- Updated Kamivision API key registration URL in documentation. - No functional or code changes; documentation only.
v1.0.0
Initial release of the Kami Image Search skill. - Capture frames from camera streams and describe them with AI for natural language search. - Import and index images manually; supports JPEG, PNG, BMP, WebP formats. - Search your visual history using text queries, with time-range filtering and duplicate frame detection. - Integrates with Kami SmartHome ecosystem and the Kamivision cloud API. - Simple setup via `setup.sh`, requiring your own Kamivision API key and camera stream configuration.
元数据
Slug kami-image-search
版本 1.0.1
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

Kami Image Search 是什么?

A skill by Kami SmartHome. Capture frames from your camera, describe them with AI, and search your visual history using natural language. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 34 次。

如何安装 Kami Image Search?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install kami-image-search」即可一键安装,无需额外配置。

Kami Image Search 是免费的吗?

是的,Kami Image Search 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Kami Image Search 支持哪些平台?

Kami Image Search 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Kami Image Search?

由 KamiVision(@13681882136)开发并维护,当前版本 v1.0.1。

💬 留言讨论