PyGOFPID

PyGOFPID is a Python package implementing
perimeter intrusion detection (PID)
systems for video protection.

Leveraging OpenCV, you can easily
define a pipeline using good old fashioned (GOF) computer vision methods, like
foreground detection and
mathematical morphology.

PyGOFPID is distributed under the open source 3-clause BSD license.

Description

Class GOFPID is a versatile tool allowing to build a parameterizable PID
pipeline, composed of:

1. input frame denoising by spatial blurring;
2. foreground detection by background subtraction or frame differencing;
3. foreground mask denoising by mathematical morphology;
4. foreground blob creation;
5. blob tracking (WIP);
6. post-filtering;
7. intrusion detection.

See examples/video.py for a simple example.

Comparison on i-LIDS

GOFPID is useful for comparison between ancient and new generations of PID,
between good old fashioned and deep learning based PID.

i-LIDS is a standard dataset
for testing video based PID.

See pyPID for a benchmark between
PID systems on i-LIDS dataset, with results given for
edge-level evaluation after 1s
(precision - recall - F1-score):

See examples/dataset_ilids.py for GOFPID configuration of i-LIDS views.

Installation

From sources

To install PyGOFPID as a standard module:

pip install git+https://github.com/qbarthelemy/PyGOFPID

To install PyGOFPID in editable / development mode, in the folder:

python setup.py develop

Testing

Use pytest.