Smart Garden

A real-time IoT monitoring dashboard that integrates multiple smart home platforms with weather data and AI-powered bird detection into a unified web interface. Designed to run 24/7 on a Raspberry Pi 5 connected to a TV, with full bilingual (English/Chinese) support.

Features

SensorPush Environmental Monitoring

Real-time data from wireless SensorPush sensors including temperature, humidity, dew point, barometric pressure, and VPD (Vapor Pressure Deficit). Interactive historical charts with selectable time ranges (1H, 6H, 24H, 3D, 7D).

Reolink Camera with HLS Streaming

Live RTSP-to-HLS video streaming from a Reolink RLC-811A camera monitoring a koi pond. Supports PTZ (Pan/Tilt/Zoom) control, snapshots, stable (sub-stream copy, 0% CPU) and HD (720p transcode) modes, with automatic reconnection and a latency watchdog.

AI-Powered Bird Detection

BirdNET-based acoustic bird detection using a local microphone. Detected species are logged with timestamps and audio clips, then enriched with AI-generated descriptions via the Claude API. Includes a leaderboard, Bird of the Day (BOTD), a searchable field guide with swipe gestures, and species activity heatmaps.

Tuya Smart Device Control

Integration with Tuya-compatible devices:

Smart Dual Water Timer -- two independently controllable valves (flower basket & koi pond) with countdown timers and server-side auto-off
Water Leak Sensor -- dual-probe pond alarm with event logging
CO Alarm -- carbon monoxide monitoring

Weather Integration

NWS (National Weather Service) forecast data including current conditions, hourly/daily forecast, and computed sunrise/sunset times.

Water Level Monitoring

Camera-based computer vision water level detection for the koi pond, with a calibration UI for ROI (Region of Interest) selection.

Animated Garden Visualization

Canvas-rendered 3D isometric garden scene with animated sprites, day/night cycle based on sunrise/sunset, and a built-in sprite editor.

Bilingual Support

All pages support English and Chinese via URL prefix (/cn/...). The TV display defaults to Chinese with Celsius.

Pages

Route	Description
`/`	Main dashboard -- weather, sensors, charts, water timer, bird notifications
`/tv`	1080p TV display -- large fonts, no emoji (for Linux/RPi5), auto-reconnecting camera
`/koi`	Koi pond camera -- HLS stream, PTZ controls, valve controls, motion events
`/bird`	Bird detection -- leaderboard, Bird of the Day, field guide with swipe gestures
`/water`	Water level -- calibration ROI tool, level history
`/edit`	Garden editor -- 3D sprite editor (dev tool)

All pages are also available in Chinese at /cn/... (e.g., /cn/tv, /cn/bird).

Setup

Prerequisites

Node.js >= 18
FFmpeg (for RTSP-to-HLS camera streaming)
Python 3 + BirdNET (for bird detection, optional)
A SensorPush account with sensors
A Tuya IoT Platform developer account
A Reolink camera (optional, for live streaming)
An Anthropic API key (optional, for AI bird descriptions)

Installation

git clone https://github.com/quake0day/sensorpush-dashboard.git
cd sensorpush-dashboard
npm install

Configuration

Create a .env file in the project root:

# ---- SensorPush ----
# Create an account at https://www.sensorpush.com
# These are your SensorPush app login credentials
SENSORPUSH_EMAIL=your-email@example.com
SENSORPUSH_PASSWORD=your-password

# ---- Tuya IoT Platform ----
# Create a project at https://platform.tuya.com
# Go to Cloud > Development > your project > Overview for these credentials
TUYA_ACCESS_ID=your-tuya-access-id
TUYA_ACCESS_SECRET=your-tuya-access-secret
TUYA_BASE_URL=https://openapi.tuyaus.com    # US data center, change for other regions

# ---- Reolink Camera (optional) ----
# Local IP address and credentials of your Reolink camera
REOLINK_IP=192.168.1.100
REOLINK_USER=admin
REOLINK_PASSWORD=your-camera-password

# ---- Anthropic / Claude API (optional) ----
# For AI-generated bird species descriptions in the field guide
# Get your key at https://console.anthropic.com
ANTHROPIC_API_KEY=sk-ant-...

# ---- Server ----
PORT=3088

Running

# Start the server
node server.js

# Or use npm
npm start

The dashboard will be available at http://localhost:3088.

Production Deployment (Raspberry Pi)

The project includes systemd service files for auto-start and auto-deploy:

# Auto-deploy checks GitHub every 60 seconds for new commits
sudo systemctl enable sensorpush-autodeploy.timer
sudo systemctl start sensorpush-autodeploy.timer

# Main service
sudo systemctl enable sensorpush-dashboard
sudo systemctl start sensorpush-dashboard

API Endpoints

Sensors

Endpoint	Description
`GET /api/sensors`	List all SensorPush sensors
`POST /api/samples`	Get historical sensor data
`GET /api/gateways`	List SensorPush gateways

Weather

Endpoint	Description
`GET /api/weather`	NWS forecast + sunrise/sunset

Camera

Endpoint	Description
`GET /api/camera/status`	HLS stream status
`POST /api/camera/stream`	Switch quality (stable/hd)
`POST /api/camera/restart`	Restart HLS stream
`GET /api/camera/snap`	Take a snapshot
`POST /api/camera/ptz`	PTZ control

Water Timer

Endpoint	Description
`GET /api/timer/status`	Valve status and countdowns
`POST /api/timer/valve`	Control valve (on/off/countdown)

Bird Detection

Endpoint	Description
`GET /api/birds/latest`	Recent detections
`GET /api/birds/daily/:date`	Detections for a specific date
`GET /api/birds/stats`	Species statistics
`GET /api/birds/botd`	Bird of the Day
`GET /api/birds/detail/:name`	AI-generated species detail
`GET /api/birds/image/:name`	AI-generated species image

Water Level

Endpoint	Description
`GET /api/water/current`	Current water level reading
`GET /api/water/readings`	Historical readings
`GET /api/water/calibration`	Get calibration config
`POST /api/water/calibration`	Set calibration ROI

Tech Stack

Backend: Node.js + Express
Frontend: Vanilla JS + HTML/CSS (no build step)
Charts: Chart.js
Streaming: FFmpeg (RTSP to HLS)
Bird Detection: BirdNET (Python) + Claude API
APIs: SensorPush, Tuya Cloud, NWS, Anthropic

License

MIT

Smart Garden 智慧花园

一个实时物联网监控面板，将多个智能家居平台与天气数据和 AI 鸟类识别整合到统一的 Web 界面中。可在树莓派 5 上 24/7 运行并连接电视显示，完整支持中英双语。

功能特性

SensorPush 环境监测

通过无线 SensorPush 传感器实时采集温度、湿度、露点、气压和 VPD（蒸气压差）数据。支持交互式历史图表，可选择时间范围（1小时、6小时、24小时、3天、7天）。

Reolink 摄像头 HLS 直播

通过 Reolink RLC-811A 摄像头对锦鲤池进行 RTSP 转 HLS 实时视频推流。支持 PTZ 云台控制、快照抓拍，提供稳定模式（子码流直传，0% CPU 占用）和高清模式（720p 转码），具备自动重连和延迟看门狗机制。

AI 鸟类识别

基于 BirdNET 的声学鸟类检测，通过本地麦克风采集音频。检测到的鸟种会记录时间戳和音频片段，并通过 Claude API 生成 AI 描述。包含排行榜、每日之鸟（BOTD）、支持滑动手势的鸟类图鉴，以及物种活动热力图。

涂鸦（Tuya）智能设备控制

集成涂鸦兼容设备：

智能双路浇水定时器 -- 两个独立控制的阀门（花篮和锦鲤池），支持倒计时和服务端自动关闭
水浸传感器 -- 双探头池塘报警，带事件日志
一氧化碳报警器 -- CO 浓度监测

天气集成

NWS（美国国家气象局）天气预报数据，包括当前天气状况、逐时/逐日预报，以及计算的日出日落时间。

水位监测

基于摄像头的计算机视觉水位检测，用于监控锦鲤池水位，配有 ROI（感兴趣区域）校准界面。

动态花园可视化

Canvas 渲染的 3D 等距花园场景，带有动画精灵图、基于日出日落的昼夜切换，以及内置精灵图编辑器。

中英双语支持

所有页面通过 URL 前缀 (/cn/...) 支持中英文切换。电视显示页面默认使用中文和摄氏度。

页面

路由	说明
`/`	主面板 -- 天气、传感器、图表、浇水定时器、鸟类通知
`/tv`	1080p 电视显示 -- 大字体、无 emoji（适配 Linux/RPi5）、摄像头自动重连
`/koi`	锦鲤池摄像头 -- HLS 直播、云台控制、阀门控制、移动侦测事件
`/bird`	鸟类识别 -- 排行榜、每日之鸟、支持滑动手势的鸟类图鉴
`/water`	水位监测 -- ROI 校准工具、水位历史
`/edit`	花园编辑器 -- 3D 精灵图编辑器（开发工具）

所有页面均可通过 /cn/... 访问中文版（如 /cn/tv、/cn/bird）。

安装部署

环境要求

Node.js >= 18
FFmpeg（用于 RTSP 转 HLS 摄像头推流）
Python 3 + BirdNET（用于鸟类检测，可选）
SensorPush 账户及传感器
涂鸦 IoT 开放平台开发者账户
Reolink 摄像头（可选，用于视频直播）
Anthropic API 密钥（可选，用于 AI 鸟类描述）

安装

git clone https://github.com/quake0day/sensorpush-dashboard.git
cd sensorpush-dashboard
npm install

配置

在项目根目录创建 .env 文件：

# ---- SensorPush ----
# 在 https://www.sensorpush.com 注册账户
# 填写你的 SensorPush 登录凭据
SENSORPUSH_EMAIL=your-email@example.com
SENSORPUSH_PASSWORD=your-password

# ---- 涂鸦 IoT 开放平台 ----
# 在 https://platform.tuya.com 创建项目
# 进入 云开发 > 你的项目 > 概览 获取以下凭据
TUYA_ACCESS_ID=your-tuya-access-id
TUYA_ACCESS_SECRET=your-tuya-access-secret
TUYA_BASE_URL=https://openapi.tuyaus.com    # 美国数据中心，其他地区请更换

# ---- Reolink 摄像头（可选）----
# 摄像头的局域网 IP 和登录凭据
REOLINK_IP=192.168.1.100
REOLINK_USER=admin
REOLINK_PASSWORD=your-camera-password

# ---- Anthropic / Claude API（可选）----
# 用于在鸟类图鉴中生成 AI 物种描述
# 在 https://console.anthropic.com 获取密钥
ANTHROPIC_API_KEY=sk-ant-...

# ---- 服务器 ----
PORT=3088

启动

# 启动服务
node server.js

# 或使用 npm
npm start

启动后访问 http://localhost:3088 即可打开面板。

生产环境部署（树莓派）

项目包含 systemd 服务文件，支持开机自启和自动部署：

# 自动部署：每 60 秒检查 GitHub 是否有新提交
sudo systemctl enable sensorpush-autodeploy.timer
sudo systemctl start sensorpush-autodeploy.timer

# 主服务
sudo systemctl enable sensorpush-dashboard
sudo systemctl start sensorpush-dashboard

API 接口

传感器

接口	说明
`GET /api/sensors`	列出所有 SensorPush 传感器
`POST /api/samples`	获取传感器历史数据
`GET /api/gateways`	列出 SensorPush 网关

天气

接口	说明
`GET /api/weather`	NWS 天气预报 + 日出日落

摄像头

接口	说明
`GET /api/camera/status`	HLS 推流状态
`POST /api/camera/stream`	切换画质（stable/hd）
`POST /api/camera/restart`	重启 HLS 推流
`GET /api/camera/snap`	抓拍快照
`POST /api/camera/ptz`	云台控制

浇水定时器

接口	说明
`GET /api/timer/status`	阀门状态和倒计时
`POST /api/timer/valve`	控制阀门（开/关/倒计时）

鸟类识别

接口	说明
`GET /api/birds/latest`	最近检测记录
`GET /api/birds/daily/:date`	指定日期的检测记录
`GET /api/birds/stats`	物种统计
`GET /api/birds/botd`	每日之鸟
`GET /api/birds/detail/:name`	AI 生成的物种详情
`GET /api/birds/image/:name`	AI 生成的物种图片

水位

接口	说明
`GET /api/water/current`	当前水位读数
`GET /api/water/readings`	历史水位数据
`GET /api/water/calibration`	获取校准配置
`POST /api/water/calibration`	设置校准 ROI

技术栈

后端: Node.js + Express
前端: 原生 JS + HTML/CSS（无需构建）
图表: Chart.js
视频推流: FFmpeg（RTSP 转 HLS）
鸟类检测: BirdNET (Python) + Claude API
API 集成: SensorPush、涂鸦云、NWS、Anthropic

许可证

MIT

Name		Name	Last commit message	Last commit date
backup		backup
garden		garden
lib		lib
public		public
screenshots		screenshots
scripts		scripts
.gitignore		.gitignore
AGENTS.md		AGENTS.md
CLAUDE.md		CLAUDE.md
README.md		README.md
bird_detector.py		bird_detector.py
blink_client.py		blink_client.py
blink_motion.py		blink_motion.py
detector_config.py		detector_config.py
motion_detector.py		motion_detector.py
package-lock.json		package-lock.json
package.json		package.json
requirements.txt		requirements.txt
server.js		server.js
test-tuya.js		test-tuya.js
water_detector.py		water_detector.py

Latest commit

History

Folders and files

Repository files navigation