Ensuring Security with Computer Vision

Architecture

In our implementation, we have used devices and computers which are general in terms of computation, performance and efficiency. The results would be more satisfactory if devices and computers with higher computational ability could be used. In the next iteration, such devices can be used. Considering computation for better computational time and efficiency, we could reduce the number of recognition performed on the same person while they are in our observation window. We can achieve this by tracking a recognized person while they are in our observation window. We have proposed and implemented a security system considering the scenario of an organizational security. Our aim was to derive data in real time so that extracted data can be helpful as a tool to ensure and enhance security.

Programming Language and modules:

Language: Python .

Modules: OpenCV, imutils, datetime, MySQLdb, Image from PIL, OS, numpy.

Database:

Database: mysql server version- 5.5.46.

Algorithms

Training:

1. Haar Cascade frontal face classifier
2. Local Binary Pattern Histogram.

Motion detecting:

1. Background subtraction
2. Dilate
3. Find Contour

Face recognition:

1. Haar Cascade frontal face classifier
2. Predict with Local Binary Pattern Histogram.