Design and implementation of a person identification system for the security of access to condominiums, based on the LBPH Faces facial recognition algorithm

Verdeguer-Valderrama D.; Campos-Vasquez N. · Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology (2021) · DOI: 10.18687/laccei2021.1.1.213

Resumen

The purpose of the work presented below is the design and implementation of a person identification system for the security of access to condominiums, based on the LBPHFaces facial recognition algorithm. Different research works were identified with respect to facial recognition for the safety of people and thus analyze the algorithmic methods they used. 300 images of the faces of the people that the system will identify were captured and stored in a repository with the name of the person's face. The algorithm identifies the faces, transforms the images to grayscale and trains them to identify if the face in real time displayed by the surveillance camera has similarity to the images trained within the repository. The 3 most used methods were analyzed, which are the FisherFaces, EigenFaces and the LBPHFaces. In the end, the results yielded very positive values with respect to the chosen algorithms, and it was concluded that the LBPHFaces was superior to the other two mentioned in response time, training and percentage of hits, which makes it the most reliable for the safety of the people.

Instituciones: Universidad Privada del Norte