Mini Raspberry Pi AI Wildlife monitor!

Video here: https://youtu.be/qhY_3XCSYsM

1. Installing Requirements!

Steps for setting up your raspberry pi! You'll need to install a few things first...

Base System Requirements

The following setup was verified on a Raspberry Pi 5 and Pi Zero 2W with Raspberry Pi OS Trixie.
It requires system wide Python>=3.13
It's best if you start with a fresh OS...

Update Pi if you haven't already

sudo apt update && sudo apt full-upgrade -y

Picamera2

Picamera2 will already be install on the full desktop version. On systems where Picamera2 is supported but not pre-installed you can install it with:

sudo apt install python3-picamera2 -y

Use this slightly reduced installation for installing on a Raspberry Pi OS Lite system!

sudo apt install python3-picamera2 --no-install-recommends -y

IMX500 (AI Camera)

sudo apt install imx500-all -y

OpenCV

picamera2 tells us to install system wide. Therefore we need to install opencv etc, also at the system level...

sudo apt install python3-opencv -y

OS Lite!

If you're using the Lite OS you will also need to install:

sudo apt install git -y

Python uv

Python packager manager uv is the preferred method for operating the mini_ai_camera.

curl -LsSf https://astral.sh/uv/install.sh | sh

Reboot Pi after install!

sudo reboot now

2. mini_ai_camera Installation

Clone the repo:

git clone https://github.com/LukeDitria/mini_ai_camera.git

Install requirements including system-wide packages (we need to use the system picamera2 install...)

cd mini_ai_camera
uv venv --system-site-packages
uv sync

Quick-start

uv run ai_cam

Install AI Camera Service

Repo comes with an install command to setup the systemd service

uv run ai_cam install

3. Updating the config.json

When you install the service a default config.json file will be created in the mini_ai_camera directory. Subsequent restarts of the service will load configuration parameters from this config.json.

You can change the behaviour of the services by editing and saving this file and restarting the AI detector services.

uv run ai_cam restart

Configuration

All settings live in config.json

Key	Default	Description
output_dir	output	Local fallback output directory
device_name	site1	Name embedded in output filenames
model	models/yolov8n.rpk	Path to the compiled yolo model file
labels	models/coco_labels.txt	Path to class labels
valid_classes	(none)	Optional path to a subset of classes to detect
confidence	0.5	Detection confidence threshold for raw detections (0–1)
iou_threshold	0.5	NMS IoU threshold (0–1)
ips	5	Max inferences per second
video_size	"1920,1080"	Camera resolution as "width,height"
buffer_secs	3	Circular video buffer length in seconds (Pre-Capture time)
ema_alpha	0.2	EMA smoothing factor for per-class confidence (lower=slower) (0–1)
event_activate	0.8	EMA threshold to trigger an active event (0–1)
event_deactivate	0.5	EMA threshold to deactivate an event (0–1)
save_video	false	Save H.264 video clips?
save_images	false	Save JPEG frames on detection?
save_data	true	Save per-detection JSON files?
draw_bbox	false	Draw bounding boxes on saved images?
auto_select_media	false	Auto-detect USB drive under /media for output?

4. More about systemd

(i) systemd is the standard system and service manager for modern Linux distributions. Once installed, you can check the status, start, stop, or restart the Ai Cam services using the systemctl command:

sudo systemctl status ai_data_logger.service

For example, to stop and disable the service so it will no longer run on boot:

sudo systemctl stop ai_data_logger.service
sudo systemctl disable ai_data_logger.service

While the status of services can be viewed with systemctl as shown above, the log output can be followed using journalctl.

To follow the live log output from the service:

journalctl -u ai_data_logger.service -f

(i) journalctl is a Linux command-line tool for viewing and managing logs from systemd. Logs can be filtered by process and time. Learn more.

5. Auto Mounting a USB Drive! (Optional)

If auto_select_media is set to true (it is false by default) the data_logger will try to find a storage device in /media to save image/video/data to.

If you are using the full desktop OS then ANY USB storage device will be automatically mounted in /media. However, if you are using the OS Lite this will not happen and you will need to configure every USB device you want to use so it will auto mount when plugged in...

📂 Auto-Mounting a USB Drive by UUID

If you want your Raspberry Pi (or Linux system) to automatically mount a USB drive at boot, you can use its UUID in /etc/fstab. This ensures the correct drive is mounted every time, even if the device path (/dev/sda1, /dev/sdb1, etc.) changes.

1. Find the UUID of Your USB Drive

First, plug in your USB drive and find its partition (e.g /dev/sda1):

lsblk -o NAME,SIZE,MODEL,MOUNTPOINT

then find it's UUID (replace /dev/sda1 with your USB device partition)

sudo blkid /dev/sda1

You'll see something like:

/dev/sda1: UUID="17F8-3814" BLOCK_SIZE="512" TYPE="vfat"

Note down the UUID and TYPE

Create a mount point

sudo mkdir -p /media/pi/myusb
sudo chown -R pi:pi /media/pi/myusb/

Edit /etc/fstab to include your device

sudo nano /etc/fstab

Add this line at the end using YOUR UUID and TYPE!!

UUID=17F8-3814  /media/pi/myusb  vfat  defaults,uid=1000,gid=1000,umask=000  0  0

You may need to run

systemctl daemon-reload

Testing that it works

sudo mount -a
df -h

You should see a line like

/dev/sda1       115G  140M  115G   1% /media/pi/myusb

Reboot!

Reboot your Pi and then run

df -h

To see if it has mounted automatically!