Installation notes

Install OS

Install latest Pi OS (tested June 25th, 2025)

Test conditions:

• Raspberry Pi 4 Model B Rev 1.5 [ 2GB ]
• Raspberry Pi OS (64-bit, released 2024-03-15) - from raspberry pi imager
• Raspbian GNU/Linux 12 (bookworm)
• Debian version: 12.0
• Kernel version: Linux 6.6.20+rpt-rpi-v8 aarch64

Setup locale, timezone, keyboard, hostname, ssh

Environment variables

Several environment variable are used for configuration/running. To accesss the AutoPollS UI, you will need to set up a custom user and password, which you will do in the bashrc script. Open Terminal on your pi and run the following command:

sudo nano ~/.bashrc

This will open the bashrc file in the nano text editor. Now add the lines to the bashrc script. NB: these must be at the TOP of your bashrc (before the 'If not running interactively'... line).

export PCAM_USER="camera login user name"
export PCAM_PASSWORD="camera login password"

After you have added these lines at the top of your bashrc script, you can save and exit by pressing CTRL+X, then selecting 'Y' (yes) when asked if you want to save the modified buffer. You will be prompted to do both of these steps in the nano editor.

Clone this repository

Prepare for and clone this repository

. ~/.bashrc
mkdir -p ~/AP
cd ~/AP
git clone https://github.com/Crall-Lab/Autopolls.git

*NB select 'y' if prompted whether you want to continue...

Install pre-requisites

sudo apt update
sudo apt install python3-numpy python3-opencv python3-requests python3-flask python3-systemd nginx-full vsftpd virtualenvwrapper apache2-utils python3-gst-1.0 gstreamer1.0-tools nmap
echo "source /usr/share/virtualenvwrapper/virtualenvwrapper.sh" >> ~/.bashrc

Setup virtualenv

. ~/.bashrc
mkvirtualenv --system-site-packages autopolls -p `which python3`
workon autopolls
echo "source ~/.virtualenvs/autopolls/bin/activate" >> ~/.bashrc

Install pandas for CSV export support and libssystemd

pip install pandas
sudo apt install libsystemd-dev

Install tfliteserve

mkdir -p ~/AP
cd ~/AP
git clone https://github.com/braingram/tfliteserve.git
cd tfliteserve

# Latest installs have required previous setuptools version
pip3 install setuptools==65.7.0

# the following tflite runtime installation instructions are from here: https://www.tensorflow.org/lite/guide/python

# install edge support
echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list
curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -
sudo apt-get update
#sudo apt install python3-tflite-runtime
pip install tflite-runtime==2.14.0
sudo apt-get install libedgetpu1-std

pip3 install -e .
# get model (TODO 404 permission denied, host this in repo or publicly)
wget https://github.com/braingram/tfliteserve/releases/download/v0.1/200123_2035_model.tar.xz
tar xvJf 200123_2035_model.tar.xz

Install this repository

cd ~/AP/Autopolls
pip install -e .
pip install uwsgi

Move latest object detection model to tfliteserve folder

sudo cp /home/pi/AP/Autopolls/tflite_20220630_1/ /home/pi/AP/tfliteserve/tflite_20220630_1 -r
sudo mkdir /home/pi/AP/tfliteserve/tflite_2023/
sudo cp /home/pi/AP/Autopolls/testModels/ssd_mobilenetV2_fpnlite_UINT8_AP24.tflite /home/pi/AP/tfliteserve/tflite_2023/ssd_single.tflite
sudo cp /home/pi/AP/Autopolls/testModels/ssd_mobilenetV2_fpnlite_UINT8_AP24_edgetpu.tflite /home/pi/AP/tfliteserve/tflite_2023/ssd_single_edge.tflite
sudo cp /home/pi/AP/Autopolls/testModels/ssd_mobilenetV2_fpnlite_UINT8_AP26.tflite /home/pi/AP/tfliteserve/tflite_2023/ssd_multi.tflite
sudo cp /home/pi/AP/Autopolls/testModels/ssd_mobilenetV2_fpnlite_UINT8_AP26_edgetpu.tflite /home/pi/AP/tfliteserve/tflite_2023/ssd_multi_edge.tflite
sudo cp /home/pi/AP/Autopolls/testModels/ssd_mobilenetV2_fpnlite.txt /home/pi/AP/tfliteserve/tflite_2023/multi.txt
sudo cp /home/pi/AP/Autopolls/testModels/efficientDetLite1_2024_tAll_edgetpu.tflite /home/pi/AP/tfliteserve/tflite_2023/EDV1_single_e.tflite
sudo cp /home/pi/AP/Autopolls/testModels/efficientDetLite1_2024_tAll.tflite /home/pi/AP/tfliteserve/tflite_2023/EDV1_single.tflite
sudo cp /home/pi/AP/Autopolls/tflite_20220630_1/labels.txt /home/pi/AP/tfliteserve/tflite_2023/single.txt

sudo cp /home/pi/AP/Autopolls/configs /home/pi/Desktop/configs
sudo cp /home/pi/AP/Autopolls/pcamPreview.py /home/pi/pcamPreview.py
sudo chmod 777 /home/pi/Desktop/configs
sudo chmod 777 /home/pi/AP/tfliteserve/tflite_2023/

Install json reading package

sudo apt-get install jq

Setup storage location

You will need a properly formatted external hard drive. These instructions will help you format your hard drive directly on the pi, but only need to be run once (i.e., if your hard drive has been previously formatted, you can skip this section). NB this will delete all existing data on your hard drive

1. Connect your external USB hard drive to the pi.
2. The software assumes you're using an external storage drive that is initally mounted at /dev/sda1. To confirm this, check the thumbdrive mounting location using - "sudo fdisk -l" in Terminal. You should see '/dev/sda1' the 'Device' output.
3. To format the drive as ntfs (the most tested format for AutoPollS, and which allows for >2TB volumes), first open 'fdisk' in Terminal using the following command:

sudo fdisk /dev/sda

4. Type 'g': this will switch to gpt
5. Type 'd': this will delete existing partitions, if any
6. Type 'n': this makes a new partion that takes up all disk space. NB use all defaults for partition number, first sector, and last sector (i.e., hit 'enter' three time)
7. Type 't' then '11' when prompted for partition type: this will switch the partion type to microsoft basic data
8. Type 'w': this writes the changes through fdisk
9. Run the following command in Terminal to make the filesystem:

# make ntfs filesystem
sudo mkfs.ntfs -f /dev/sda1

Mount storage location

echo "/dev/sda1 /mnt/data auto defaults,nofail,user,uid=1000,gid=124,umask=002  0 0" | sudo tee -a /etc/fstab
sudo mkdir /mnt/data
sudo mount /mnt/data
sudo mkdir -p /mnt/data/logs
sudo chown pi /mnt/data
sudo chgrp ftp /mnt/data
sudo chmod 775 /mnt/data

Change hostname file permissions

sudo chmod 777 /etc/hostname

Setup web server (for UI)

sudo htpasswd -bc /etc/apache2/.htpasswd pcam $PCAM_PASSWORD
sudo rm /etc/nginx/sites-enabled/default
sudo ln -s /home/pi/AP/Autopolls/services/pcam-ui.nginx /etc/nginx/sites-enabled/

Setup systemd services

This will set up the systemd services to run the AutoPollS software in the background. NB: the overview service and timer are not needed for usb cameras, and may be removed in a future update

cd ~/AP/Autopolls/services
for S in \
    tfliteserve.service \
    pcam-discover.service \
    pcam-overview.service \
    pcam-overview.timer \
    pcam@.service \
    pcam-ui.service; do \
  sudo ln -s ~/AP/Autopolls/services/$S /etc/systemd/system/$S
done
# enable services to run on boot
for S in \
    tfliteserve.service \
    pcam-discover.service \
    pcam-overview.timer \
    pcam-ui.service; do \
  sudo systemctl enable $S
done
# start services
for S in \
    tfliteserve.service \
    pcam-discover.service \
    pcam-ui.service; do \
  sudo systemctl start $S
done
sudo systemctl restart nginx

Add script to fix and remount /dev/sda1 if corrupted

We find that there is some fairly frequent ntfs corruption of the mounted drives when plugged in the Autopolls system and running over long periods. The following scripts should look for this file error and fix it.

In terminal open crontab, via "crontab -e", at the bottom of the file include:

* * * * * sudo python3 /home/pi/AP/Autopolls/mountFix.py

Install MCC134 libraries and script

Attach the MCC134 thermocouple ontop of the Pi's 40 pin GPIO, then run commands below

cd ~
sudo apt-get install libraspberrypi-dev raspberrypi-kernel-headers

cd ~
git clone https://github.com/mccdaq/daqhats.git
cd ~/daqhats
sudo ./install.sh

If libraries do not install automatically

~/.virtualenvs/autopolls/bin/pip install daqhats
deactivate
sudo pip install daqhats --break-system-packages

sudo chmod 775 ~/AP/Autopolls/tempSensor.py
sudo mv ~/AP/Autopolls/tempSensor.py ~/daqhats/examples/python/mcc134/tempSensor.py

Open crontab ('crontab -e' in Terminal) and add this line to the bottom of the script:

* * * * * sudo python3 ~/daqhats/examples/python/mcc134/tempSensor.py

Test that the temperature probes are reading correctly. First, reboot the pi, which you can do with the following command:

sudo reboot -h now

sudo python ~/daqhats/examples/python/mcc134/tempSensor.py

Then confirm a folder in /mnt/data titled "tempProbes", which should contain csv with a temp reading recorded into a csv file. The csv file should contain the hostname and date.

Install wittyPi libraries and script

Attach the wittyPi on top of the thermocouples 40 pin GPIO, then run commands below. If using an old wittyPi replace with "WittyPi3"

wget http://www.uugear.com/repo/WittyPi4/install.sh
sudo sh install.sh

Reboot your pi again (manually, or using 'sudo reboot -h now'), then run the following command in Terminal:

sudo mv ~/AP/Autopolls/schedule.wpi ~/wittypi/schedule.wpi
sudo ~/wittypi/runScript.sh

Viewing cameras option 1: camera view and focus

To view the four cameras and check orientation and focus, run the following command in terminal:

python ~/AP/Autopolls/pcamPreview.py -t 30

This will give you a (30-second) preview of the cameras you have attached. You can adjust the '-t' parameters to change the length of the preview. To test focal distance values, you can use the optional -f input value. If not value for -f is input, the script will default to pulling the focal distance specified in the config file on the Desktop.

Viewing camera option 2: web browser

In the browser on the pi, you can view attached cameras and change parameters by connecting to the UI: open a browser, and type in "127.0.0.1". FYI, you will have to enter. This provide an overview of recent detections, but does not provide a realtime view of the cameras

If errors with systemD

pip3 install systemd

troubleshooting commands via systemctl

# check that model is loaded and executed correctly
sudo systemctl status tfliteserve.service

# check that each camera is running correctly, hit tab to complete command for each camera
sudo systemctl status pcam@