Image classification and object detection, applied to a public webcam in the city of Münster.
See webcam data source here.
- Tested on
Ubuntu 20.04 ffmpeg, e.g. on Ubuntu install with:
sudo apt install ffmpeg-
Create and activate virtual environment. For example with
conda:conda env create -n cam python=3.7 conda activate cam
-
Install Python dependencies and
traffic-campackage (with virtual environment activated):pip install -r requirements.txt pip install -e . -
Run scripts in
bin/in their numerical order. This includes downloading images, training a classifier, and counting persons.