test-image.py

Code taken from https://www.pyimagesearch.com/2015/03/30/accessing-the-raspberry-pi-camera-with-opencv-and-python/.

From a machine with an X server:

ssh -X pi@<IP of Pi>

Then on the Pi:

cd opencv
make run-container

Then in the container:

python test-image.py

Pi setup

wpa_supplicant.conf with passwords for my home network and phone hotspot.

Camera enabled (raspi-config -> /boot/config.txt).

sudo aptitude install emacs git docker.io imagemagick virtualenv python-dev

(.emacs and org repo copied from work laptop)

(Emacs packages installed: paredit, auto-complete, go-autocomplete, go-guru, magit.)

git clone https://github.com/neiljerram/opencv.git

opencv/Makefile has run-container target that runs shell in a container that maps the opencv subdir and adds in an OpenCV install.

Unnecessary venv

I thought I needed a venv at opencv/venv, and in that venv, pip install picamera[array].

I now think that is unnecessary, because it appears that the sgtwilko/rpi-raspbian-opencv:latest image already includes picamera (which makes sense given the Pi focus); and because I’m not doing anything to activate the venv inside the container, but code including import picamera still works.

nelljerram/opencv

test-image.py

Pi setup

Unnecessary venv