BirdWatchAI Outdoor bird feeder with camera and AI Species detection with community feed (No subscriptions) print and build it all yourself (or buy it) includes software. NO SUBSCRIPTIONS! Your data stored locally, not in cloud (unless you want it there).
Obliczymy ostateczną cenę na podstawie wybranego pliku. Sprawdzisz ją i zaakceptujesz przed płatnością.
Zapisane na tym urządzeniu.
Feeder for TapoC113 camera. Go to www.BirdwatchAI.com for all details on AI species detection DIY.
- Ocena
- 0 / 5
- Polubienia
- 5
- Pobrania
- 4
- Wydruki
- 0
- Wyświetlenia
- 60
- Oceny
- 0
Aktualizacja u źródła:
Informacje o modelu
- Źródło
- Printables
- Kategoria
- Outdoor & Garden
- Licencja
- Creative Commons — Attribution
See www.BirdWatchAI.com for all of the details on how to setup the software. Free download. You can print/build it all yourself, or purchase a setup kit.
See the live community feed of feeders currently running the BirdwatchAI software at https://www.birdwatchai.com/docs/community.html
* Updated Rev2: Made Feeder Base printable without supports (New Teardrop design for camera opening).
Bird Feeder for the TP-Link Tapo C113 — Built for AI Bird ID
A remix of Deltaken's excellent "Bird feeder camera (Wyze, Blink)", reworked so it holds a TP-Link Tapo C113 instead. All credit for the original feeder design goes to Deltaken — I recut the camera pocket and did some slight modifications listed above. Free to download, print, and remix further.
I make BirdWatchAI, software that watches a camera stream, notices when a bird lands, and tells you which species it is. Getting good identifications depends as much on the physical setup as on the software: the bird needs to be close enough, lit from the right side, and against a background that isn't moving. No amount of AI fixes a photo where the bird is 40 pixels wide.
Deltaken's design already puts the camera exactly where it needs to be, prints without supports, and snaps together in two parts. It just wasn't cut for the camera I recommend to my users — so I adapted it, and I'm posting the result here for anyone who wants it.
Why the Tapo C113
The C113 is the camera I point people toward, and it's what this version is dimensioned for:
~$30-$40 — cheap enough that you're not nervous about hanging it in the weather
2K / 3MP (2304 × 1296) — plenty of resolution for identifiable plumage detail
IP65 weatherproof — rated for outdoor use straight out of the box
Open RTSP stream — no subscription, no cloud dependency; any software can read it
USB-C power, 2 m cable — easy to extend or route
Local microSD recording — optional, and independent of whatever software you run
Body: 61.1 × 60.5 × 92.1 mm, 90 g — small and light, so nothing sags or overbalances
The camera pocket is toleranced to that body and the camera's own swivel base is removed. Other Tapo bullet cameras are similar but not identical — check before you print if you're using something else. If you have a Wyze or Blink camera instead, go grab Deltaken's original; it's the better file for you.
The macro lens is worth it
The pocket is also sized to support a strap-on macro lens on the front of the camera. The C113 is fixed-focus, and at feeder distances a macro lens is what takes a bird from "recognizable" to genuinely sharp — feather detail, eye rings, bill shape, the things that separate two similar species. If you want the best identifications this setup can produce, add the lens.
I sell the matching lens on my site, or you can pick one up on Amazon as well. The pocket fits it either way; nothing to modify.
Print settings
Material: PETG. That's what I printed mine in, specifically to handle the outdoor heat — this thing sits in the sun all day and PETG holds its shape. ASA works too if that's what you run. PLA will get you through a season in a shaded spot, but it will soften and sag in direct summer sun and go brittle with UV, so treat it as a prototype material here.
Supports: I printed with supports in center of the camera opening to hold up the arch. You can print it without slicer supports either way — you'll just get a bit of stringing across the opening if you skip them, which cleans up easily with flush cutters or a quick pass with a hobby knife. Nothing structural depends on it.
Layer height: 0.2 mm
Infill: [15]%
Orientation: Roof (top facing down), Base (top facing up)
Color: I printed mine with the base black and the rest white. I find the seed tray tends to get dirty and the black hides the dirt from the camera.
Assembly
Nothing difficult here, just snap the roof in place when filled with seeds.
Camera base is removed. Camera slides in from the front, power cable from the back through small opening. This keeps the camera centered and straight.
Mounting and framing
This is the part that makes or breaks detection quality, so it's worth a couple of minutes.
Put the sun behind the camera, not behind the bird. Backlighting silhouettes your subject and ruins otherwise perfect detections.
Keep the background boring. A fence, a wall, dense foliage — anything static. Swaying branches and moving sky trigger endless false motion events.
Check your Wi-Fi at the mounting spot before you commit. A weak signal shows up as a stuttering RTSP stream, which shows up as missed birds.
Mounting options: [pole / rail / get creative]. There are three holes in the back to mount it.
Getting the camera streaming
Whatever software you point at it, the C113 needs two things set up first:
Create a Camera Account in the Tapo app (Device Settings → Advanced Settings → Camera Account). This is a separate username and password from your Tapo login — 6–32 characters each. RTSP won't work until it exists.
Reserve a static IP for the camera in your router's DHCP settings. Skip this and the stream will mysteriously die a week later when the lease changes.
Your stream URL is then:
rtsp://USERNAME:PASSWORD@CAMERA_IP:554/stream2The software side (this is the honest pitch)
The feeder is free and always will be, and it works fine as a plain camera feeder with whatever software you like. But this version exists because it's the front end for BirdWatchAI, which is what I'd love you to try.
It watches the RTSP stream, and when something lands it captures a snapshot and a short clip, identifies the species, and notifies you.
Runs on Windows — or, better, on a Raspberry Pi. A Pi sitting next to your router, drawing a few watts, watching the feeder around the clock is the setup I'd recommend to anyone: no PC left on, nothing in the cloud, and it just runs. macOS and Linux work too.
Species ID for 965 North American species
100% local processing — identification happens on your own hardware. Nothing is uploaded. Community sharing exists, but it's opt-in and anonymized.
Notifications by email, mobile push (Pushover / NTFY), desktop toast, or straight to a digital photo frame
Regional awareness — flags how rare a species is for your location and season, which is how you find out about the interesting thing that showed up while you were at work
Snapshots, clips, weekly and monthly stats, and around 140 badges if you're the completionist type
Light on resources — a few hundred MB of RAM, and only a modest CPU bite while it's actually looking at a bird
No subscription. One-time license, and free to try for 30 days — no credit card to start. No other smart bird feeder is sold without a subscription.
Your data is stored locally (Unless you contribute it to the community site!)
Two ways to get there:
Build your own — print this feeder, buy a C113 for about $30, add the macro lens, and purchase just the software license from my site. That's the cheapest route, and it's why this file is free.
Buy the complete package — I sell ready-to-go kits at birdwatchai.com with the feeder, camera, lens, and software together all setup, for people who'd rather not source the parts and prefer just to have it all built for them!
→ birdwatchai.com — 30-day trial, kits, macro lenses, docs, and setup guides
If you print this and never buy the software, that's completely fine. I'd still like to see your build.
Keeping it working (and the birds healthy)
Clean it regularly — every couple of weeks, more often in wet weather. Wet seed grows mold, and mold makes birds sick. Warm soapy water, rinse, dry fully.
Watch the drainage — Clear blocked holes when you clean.
Squirrels will find this. You can train the software to detect the squirrels!
Expect a settling-in period. Birds may take a few days to a couple of weeks to trust a new feeder, especially one with a camera stuck to it. Black oil sunflower seed is the fastest way to build a crowd.
Wipe the lens whenever you clean the feeder — the macro lens especially. Pollen, dust, and spider webs across the glass degrade identification more than you'd expect.
Credit and remixing
The original feeder is Bird feeder camera (Wyze, Blink) by Deltaken — if you like this, go give them a like and a print. This remix is shared under Creative Commons (4.0 International License).
Remix it further, scale it, adapt the pocket for a different camera — go ahead. If something doesn't fit, prints badly, or you found a better camera angle, please say so in the comments. Photos of your build and your first good bird capture are the best thing you can leave here.
Happy printing, and happy birding.
Tags: birdfeeder, bird, birds, birdwatching, tapo, c113, camera, camera-mount, macro-lens, remix, ai, raspberry-pi, garden, outdoor, wildlife, feeder, tp-link, rtsp, petg, functional