RA-opencv

Augmented Reality with OpenCV using Python.

Usage

• Place the image of the surface to be tracked inside the references folder;
• Replace TRAIN_IMAGE_RELATIVE_PATH with the relative path of the train image you want to track;
• Replace OBJ_MODEL_RELATIVE_PATH with the relative path of the model you want to render. To change the size of the rendered model change the scale parameter MODEL_OBJ_SCALE to a suitable number. This might require some trial and error.

Running

Open a terminal session inside the project folder and run:

# Activate virtualenv
virtualenv python_ra
source python_ra/bin/activate
# to deactivate: run `deactivate`

Then:

# Install from the requirements
pip install -r requirements.txt
python3 src/ra.py

Command line arguments

• --rectangle, -r: Draws the projection of the reference surface on the video frame as a blue rectangle;
• --matches, -ma: Draws matches between reference surface and video frame.

python3 src/ra.py -r -ma

Troubleshooting

If you get the message:

Unable to capture video

printed to your terminal, the most likely cause is that your OpenCV installation has been compiled without FFMPEG support. Pre-built OpenCV packages such as the ones downloaded via pip are not compiled with FFMPEG support, which means that you have to build it manually.