Project: Track backyard birds on a digital counter

I’ve been going deep down the BirdNET-Pi rabbit hole over the last month. A BirdNET-Pi setup uses a Raspberry Pi computer and microphone to listen, ID, and log birds it hears 24/7.

On the BirdNET-Pi interface, you get lots of great daily and historical detection data, including the number of detections in the last 24 hours.

I came up with an idea to automatically show this count on some kind of digital counter. With this project idea in mind, I did research to find suitable parts to build and program it.

Contents

• Components
• Prerequisites
• Getting started
• Flashing the microcontroller
  • Wiring the display
• Configuring wledPixel
• Configuring a NodeRed flow to transmit the detection count
• Optional: Display the latest species detected
• Optional: Get the count for the current day only

Components

A heads up: this is a complicated project that requires several systems to be set up and working already.

Prerequisites

Getting started

BirdNet-Pi and BirdWeather

My BirdNET-Pi installation has been running since December 2022 and has picked up all kinds of great detections and recordings (30,000+). The station logs all of the data locally, but you can optionally submit your detections to a service called BirdWeather. BirdWeather’s API makes this whole project possible as it lets you request the data needed for the display.

Flash the microcontroller

The brains of the sign is the Wemos D1 Mini. This small ESP8266 board will be flashed with a program called wledPixel which will receive and display the messages. The process is a bit involved, so rather than re-explain it here, there are a few guides you can follow.

I used two guides to flash the D1 Mini chip. To flash the chip you’ll need to install ESPTool on your computer. I have a Mac and used this guide. Use ESPTool to flash the device using a few terminal commands from this guide. However, don’t follow it to the end and install Tasmota. (what that guide is talking about). Instead, download the wledPixel.bin file and put it in the same folder you installed ESPTool in.

Flash the chip using the commands in the second guide but replace the name of the .bin file in the commands with the wledPixel.bin name. In my case, I renamed the .bin file to wledPixel.bin to make it easier.

Once the flash is successful, unplug and replug the D1 Mini and let it boot. Connect to the Wi-fi network it is broadcasting and enter your network details to get it connected to your network. wledPixel is now ready to be configured, but first, we’ll need to wire the display to the board.

Wire the display to the board

The wledPixel GitHub page has a wiring diagram for the display and board. I don’t know how to read these confidently, so I followed along with this build video on YouTube to connect and solder the wires correctly.

Power it up (with 5V)

It’s crucial that when you connect this to an outlet for power, you use a 5V power block. Anything more will fry the D1 Mini you worked so hard on.

When you boot it up with the display connected, hopefully, it lights up and displays the device’s IP address. Navigate to this IP address in your browser and you’ll see the interface and can configure the settings.

Configure wledPixel

Feel free to play around with the settings in the configuration to get it to your liking.

Display settings

Navigate to wledPixel’s settings page and set the options as follows:

Zone0 Settings

Set Zone0’s options as follows:

MQTT Settings

Add in the info for your MQTT server.

This finishes the configuration of wledPixel.

Creating a NodeRed flow to transmit bird detections to the sign

The sign is now ready to receive the bird count. To send the count to the sign we will use NodeRed.

A NodeRed Client and MQTT server are required for this project. Since I already had a Home Assistant server in my home, I used that for the NodeRed and MQTT Server capabilities. You can use a Raspberry Pi or Libre Computer and install MQTT and NodeRed on it without having to use HomeAssistant.

A NodeRed “Flow” will be used to request the detection data, parse the response, and transmit the count in a message via the MQTT topic “birdcounter” to the digital sign.

Create a new NodeRed Flow

On the Flows page in NodeRed there may be a default “Flow 1” already there for you to begin working on. If not create a workflow and name it something like “Bird Counter”.

Add a node to request BirdWeather data

On the left side of the page are all the available nodes you can add to your flow. Search for “http”, drag an “http request” node onto your canvas, and double-click it to open up the configuration.

BirdWeather offers an API to request a station’s detection data. We’ll use this API to get the day’s data for our station.

In the configuration panel, add the following URL to the URL field, replacing [token] with the token/station ID you are provided when you get set up on BirdWeather.

https://app.birdweather.com/api/v1/stations/[token]/stats

For the “Return” option, select “a parsed JSON object”, give the node a name in the final field, then click done.

Add a node to include the detection count in your message

On the left side, search for “Change” and drag this node on your canvas next to the first node.

If you click the anchor points on the nodes and drag to another node, you’ll connect the two and it will pass along its payload (message). Once you’ve connected them, double-click the change node to open the configuration panel.

Give it a useful name and set the rules as follows. Click the icon on the second line and change it from .msg to “expression” then set the rule:

Set msg.payload to payload.detections

This adds the number of detections to the message about to be sent to the counter.

Add a node to send the final message to the counter

Now, we need a node that will transmit our final message payload to the counter which is listening on the MQTT topic of “wledPixel-B779/zone0/birdcounter”. The first segment “wledPixel-B779” will be replaced with the name of your display (see below).

On the left, search for “MQTT” and add in an “mqtt out” node and connect it to the previous node.

On the configuration panel, add the topic “wledPixel-B779/zone0/birdcounter” and a descriptive name then click “Done”

Make it loop

Next, we want to make the whole thing loop to update the number, but not constantly. Following similar steps, add a Delay node and connect it to the Extract daily detections node and, using the other side, connect it back to the first node as shown below. Using its configuration panel, set the delay to 60 seconds, leaving the other options default. This will loop the whole flow every minute.

Add a button node to start the whole thing

Add an “Inject” node to the workflow on the left side and connect it to the first node we made. When the button on this node is clicked it will start the process and it will keep running indefinitely thanks to our loop.

Click deploy and click push the virtual button

Click the red Deploy button to save and publish your new NodeRed workflow. Now is the moment of truth. If your digital sign/D1 Mini is plugged in and online listening to the “wledPixel-B779/zone0/birdcounter” topic, as soon as you click the button in your NodeRed workflow the current bird count will light up on the screen!

Optional: Display the latest species

Now, if you want to get fancy, you can display the latest species following the current bird detection count. Let’s see how to do that.

Request the latest species from BirdWeather

Remove the final loop back from above and add a new http request node.

Use another http request node and use the following URL:

https://app.birdweather.com/api/v1/stations/[token]/detections/?limit=1

Extract the species name from the returned data

The above API request will return all kinds of data about the most recent detection, including the common name of the bird. We’ll extract the common name from the data with the Change node.

Add another Change node and add the following rule:

Set msg.payload to expression: payload.detections.species.commonName

Create a template message to send to the sign

Instead of just pushing only the name to our sign, we’ll format the message as follows:

Latest bird : [Bird Name]

To accomplish this, we’ll use the Template node. Add this node to your flow and add this template:

Latest bird : {{payload}}

This updates the message and it’s ready to be sent to the sign.

Send the message

Similar to the detection count, we’ll use the mqtt out node to send the final message.

Connect the new loop

Similar to the above, create a new loop back to the beginning with a Delay node in the middle.

The final NodeRed flow. This sends both the detection count and the latest species name to the sign.

3D-printed enclosure

Using the STL file linked above, you may print a 3D enclosure for the display, which will give it a nice finished look.

Optional: Get count for current day only

The above BirdWeather API for the count gets a rolling 24-hour period of counts, so it never really goes down to zero. I wanted to have the count for the current day only so it was more fun to watch, the current dev for the project was kind enough to provide custom code to make this possible.

To accomplish this, we need to do three things. Add a custom PHP file to our BirdNet-Pi station and edit the API Request node for the count data, and edit the Change node.

Edit Count Request node

We’ll start with the node first. Edit the first node we created “BirdWeather Count Request”, then use the following URL and click save.

https://[address to your BirdNET-Pi]/today_count.php

Edit Change node

Edit the change now and use the following template:

Set msg.payload to payload.today_count

Add custom file to BirdPi-NET install

SSH into your BirdNET-Pi and run the following command. This command will download the needed file and place it in the proper location.

wget https://gist.githubusercontent.com/ehpersonal38/a34ce000e1645bc85aba936a5f89ac72/raw/43311739c3176f1b77e298ae5fabd58afc27250b/today_count.php -O ~/BirdNET-Pi/scripts/today_count.php

Now when the NodeRed flow is updated, deployed, and initiated, the flow will call on your station for the data for the current count and it will be for the current day only.