Table of Contents |
---|
Overview
Data Image collection is done on any computer running Linux on the drone. Data Image collection accesses sensors the camera in the same manner as the airside system.
Repository:
...
https://github.com/UWARG/model-training
Software
Usage
Collection:
Software
Setup - USB Camera
Follow the instructions to clone the repository and activate the environment: Autonomy Workflow Software
Install packages:
Code Block |
---|
pip install -r requirements.txt |
Usage - USB Camera
Note |
---|
Prerequisites:
|
Update the repository by running:
Code Block |
---|
git checkout main
git pull
git submodule update --remote --merge |
Copy the contents of rc.local.copyme
to /etc/rc.local
.
Modifying
/etc/rc.local
requires superuser access (e.g.sudo
)
Collecting images:
Connect the camera to the computer
Power on the computer:
If sensors are camera is not connected, data image collection fails to start and . Power off the computer needs to be restartedData and go to step 1
Image collection is automatically run on startup through the
/etc/rc.local
file
...
Collect images by pointing the camera towards the object(s) of interest
Power off the computer
Accessing images:
Disconnect all sensors and power
Power on the computer
Move or copy the data images to be saved
Delete the data images to free space for future collection
Setup
...
Code
Open the terminal and navigate to the home directory:
cd ~/
Follow the instructions: Python Repository
Install packages: pip install imaging/requirements.txt
Do not use the other requirements files
Startup
The contents of imaging/rc.local.copyme
assume the username is warg
.
OpenCV Python:
...
- Raspberry Pi Camera Module 3
clone the repository or update it as normal (same as USB Camera)
Note |
---|
DO NOT activate a virtual environment: you actually want to deactivate it before you run this script. The picamera2 library is pre-installed on PiOS, and it is not encouraged to install it using pip: https://datasheets.raspberrypi.com/camera/picamera2-manual.pdf. Hence, we will use the pre-installed library instead. Those libraries cannot be easily installed using pip and require additional steps, so we just use the provided library in global space. There are some missing dependencies when installing through pip, which causes build fail. |
Usage - Raspberry Pi Camera Module 3
Option A: Run the script manually
Code Block |
---|
python picam.py |
Option B: Run the script from startup using /etc/rc.local
.
Add the following line to /etc/rc.local
...
FFmpeg video:
...
Install FFmpeg: sudo apt-get install ffmpeg
...
(starts the image capturing in the background):
Make sure you don’t activate any virtual environments beforehand!
Code Block |
---|
cd /home/warg/image-collection
python picam.py & |