Python algorithm for traffic videos processing in VCA and regular cameras. BG/FG combined with SURF features to measure velocity. Camera georeferencing and mapping pixel coordinates in the video to coordinates in a map. Camera (VCA and PTZ) comunications with Onvif. This proyect is still in development, not organized or fully working yet.

We chose the same license as OpenCV: BSD 3-Clause License.

Group members: Agustín Yabo, Damián Estanganelli, Damián Oliva, Lilian García y Sebastián Arroyo.

To use this python code there is no installation, just copy the folder visionUNQ to your computer and import it as a package. Porting from python2 to python3 and OpenCV(2.4?).

• visionUNQ : is a packaje containing functions, objects, etc.
• examples : has working scripts accompanied by video files if necesary.
• dev : contains scrips and pices of code in development/test also mains that run using visionUNQ but haven't yet been fully tested of given definite form.
• resources : contains all non code files (images, videos, etc).
• README.md : this file
• LICENSE.md : The license that applies to all the code in the repository. BSD 3-Clause license, same as OpenCV.
• HISTORIAL.md : Log of notable changes and annotations in general.

This repository is meant to unify the python codes of people working in computer vision applied to traffic monitoring in Universidad Nacional de Quilmes (Buenos Aires, Argentina), Departamento de Ciencia y Tecnología, IACI. The idea is to organize the algorithms into a single library for easier collaboration.

Stanganelli, D., Oliva, D. E., Noblia, M., & Safar, F. (2014, June). Calibración de una cámara fisheye comercial con el modelo unificado para la observación de objetos múltiples. In Biennial Congress of Argentina (ARGENCON), 2014 IEEE (pp. 147-152). IEEE.

Procedeeng de la 2015 XVI Workshop on Information Processing and Control (RPIC) en IEEE Xplore

Tesis Yabo

Tesis Stanganelli (proximamente)

Python libraries it depends on: opencv, numpy, onvif, wxPython, pickle, PyQt, neurolab, urllib3. Most of them installed via pip if not already present.

wxPython installed from source (http://wxpython.org/Phoenix/snapshot-builds/): $ sudo python3 setup.py install