【文章标题】：I turned my security cameras into an automatic bird identification system
【文章标题】：我把家庭监控摄像头改造成了自动鸟类识别系统
【文章正文】：
My wife loves listening to the birds and identifying them using an app on her phone. I thought I’d take this a step further and see if there was something that can identify the birds based on their song.
我妻子喜欢听鸟鸣并用手机APP识别鸟类。我想更进一步，看看能否通过鸟叫声实现自动识别。
Looking around I found BirdNet and BirdNet-Go, then discovered you can run this on Docker and use the security cameras you already have outside to identify the birds. Awesome!
经过搜索，我发现了BirdNet和BirdNet-Go，更惊喜的是可以通过Docker部署，直接利用现有的户外监控摄像头来识别鸟类。
So I took 3 of the cameras around my house and used their microphones to identify the birds. It was magic.
于是我用房屋周围的3个摄像头，通过它们的麦克风实现了鸟类识别，效果神奇。
Real-Time Bird and Bat Audio Detection
实时鸟类与蝙蝠音频检测
BirdNet-Go runs 24/7. It listens constantly. The moment a bird starts singing, it analyzes the audio and gives you an identification. No waiting, no manual recording.
BirdNet-Go全天候运行，持续监听环境声音。一旦有鸟鸣响起，系统会立即分析音频并给出识别结果，无需等待或手动录制。
It works for bats too, which is pretty cool if you’ve got them flying around at dusk. The system just keeps running in the background, cataloging everything it hears. It also started detecting frogs which is interesting.
对蝙蝠也同样有效——如果你家黄昏时有蝙蝠盘旋会很有趣。系统在后台持续运行，自动归档所有识别记录，后来甚至还能检测蛙鸣。
Multi-Model Local AI Inference Engine
多模型本地AI推理引擎
The whole thing runs locally on your hardware. No cloud services. No API calls. The AI models live right on your server or Raspberry Pi.
整套系统完全在本地硬件运行，无需云服务或API调用，AI模型直接部署在你的服务器或树莓派上。
This means it’s fast, private, and doesn’t cost you anything per month. You can even run multiple models if you want different detection strategies or regional bird databases.
这意味着响应迅速、隐私安全且零月费。你还可以运行多个模型，适配不同识别策略或地区鸟类数据库。
They recently added Google Perch v2 to the model gallery allowing for 14,795 species to be detected vs the 6,000 that BirdNET 2.4 provided.
最近新增的Google Perch v2模型可识别14,795个物种，远超BirdNET 2.4的6,000种。
Alert Rules with Species List Matching
基于物种清单的警报规则
You can set up rules for specific birds. Want to know the instant a cardinal shows up? Done. Looking for a rare species in your area? Set an alert.
可针对特定鸟类设置规则：北美红雀现身时立即通知？搞定！寻找本地稀有物种？设置警报即可。
The system matches against species lists and sends you notifications based on whatever criteria you set. It’s like having a birding buddy who never sleeps.
系统会根据预设条件比对物种清单并发送通知，就像有个永不休息的观鸟伙伴。
I even have it connected to a channel in my homes Discord server. Wait, your house doesn’t have a dedicated Discord?
我甚至将其接入家庭Discord服务器的专属频道。什么？你家还没搭建专属Discord？
Species Novelty Tracking for New Detections
新发现物种追踪功能
This feature tracks which birds are new to your yard. First time a blue jay visits? BirdNet-Go flags it.
该功能会标记庭院首次出现的鸟类。冠蓝鸦初次造访？BirdNet-Go立即标注。
It maintains a running list of every species it’s detected, so you can see your yard’s biodiversity grow over time. Honestly, this turned into a fun game for us.
系统持续更新已检测物种清单，让你见证庭院生物多样性的增长。说实话，这成了我们的趣味游戏。
RTSP Stream Support for IP Cameras
支持IP摄像头的RTSP流
If your cameras support RTSP, you’re good to go. Most modern IP cameras do. You just point BirdNet-Go at the stream URL.
只要摄像头支持RTSP协议（多数现代IP摄像头都支持），输入视频流URL即可快速接入。
I used three of my existing security cameras. Didn’t need to buy any special hardware. The cameras I already had for security now pull double duty identifying birds.
我直接复用三个现有安防摄像头，无需额外硬件。原本用于监控的设备现在兼职做鸟类识别。
BirdWeather Integration for Data Sharing
BirdWeather生态数据平台集成
BirdWeather is a community platform for sharing bird detection data. BirdNet-Go integrates with it directly.
BirdWeather是鸟类观测数据共享社区平台，BirdNet-Go可直接对接。
If you want to contribute your observations to a larger dataset, you can. It helps researchers and other birders see what’s happening in your area.
你可以选择将观测数据贡献给更庞大的数据库，助力研究人员和观鸟爱好者了解本地生态。
Self-Hosted with No Cloud Dependencies
完全自托管·零云依赖
Everything lives on your network. The audio never leaves your house unless you explicitly share it.
所有数据都在本地网络处理，除非主动分享，音频资料绝不会离开你的住宅。
No subscription fees. No terms of service changes. No company shutting down the service in two years. You own the whole stack.
无订阅费、无服务条款变更、不怕企业突然停服，整套系统完全自主掌控。
It runs in Docker, so it’s easy to manage alongside everything else in your homelab.
基于Docker的运行方式，可轻松与家庭实验室的其他服务协同管理。
Can connect to Home Assistant
支持Home Assistant智能家居平台
Evernything needs to be able to connect to Home Assistant and BirdNET-Go can use MQTT to be discovered in Home Assistant.
通过MQTT协议接入Home Assistant，实现智能家居生态互联。
Visual Audio Channel Energy Level Analysis
可视化音频通道能量分析
The interface shows you real-time audio levels for each channel. You can see exactly what the microphones are picking up.
界面实时显示各通道音频能量值，清晰掌握麦克风采集状态。
This helps you position cameras better or troubleshoot why one isn’t detecting anything. Sometimes you’ll realize the camera is pointed at a noisy AC unit or getting wind noise, and you can adjust accordingly.
便于优化摄像头布设位置或排查故障——比如发现某摄像头正对嘈杂空调或受风噪干扰时，可及时调整。
Final Thoughts
最终感想
Look, this project surprised me. I expected it to be neat but figured I’d get bored after a week.
这个项目完全超出预期——原以为只是有趣的小玩意，一周后就会厌倦。
Instead, my wife and I check it daily. We’ve learned which birds visit at what times. We’ve spotted species we didn’t know lived nearby.
结果我和妻子每天查看，掌握了不同鸟类的造访时间，还发现了许多未知的邻近物种。
My friends got interested too, I made this avaiable on the public house domain name (behind Cloudflare of course.) and allowed a few of my friends to connect to it and see how it works.
朋友们也产生兴趣，我将其部署到家庭公网域名（当然经过Cloudflare保护），供他们体验运作原理。
It turned our security camera setup into something genuinely useful beyond security.
这套方案让安防摄像头实现了远超安全监控的价值。
If you’ve got cameras with microphones and even a passing interest in birds, give BirdNet-Go a shot.
如果你有带麦克风的摄像头，哪怕对观鸟只有短暂兴趣，都值得尝试BirdNet-Go。
It’s one of those homelab projects that actually improves daily life instead of just being technically interesting.
这是少数真正提升生活品质的家庭实验室项目，而非仅具技术趣味性。