Facial Attribute Recognition and its Application in Drug Abuse Detection
Date
2019-07-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Face attribute analysis is a valuable aide in biometric-based human identification. This is a challenging task due to variations in lighting, occlusion, pose and other variables. This work proposes an effective and robust approach to detect up to 40 face attributes using deep machine learning models such as Convolutional Neural Networks(CNNs). The focus is on using different pretrained CNNs to extract features from the intermediate CNN layers for face attribute recognition. Also, feature level fusion is proposed by concatenating the features extracted from the intermediate layers of the CNNs, thus using ensemble features. Classification of face attributes was performed
using a linear Support Vector Machine (SVM) and end-to-end training of using CNN for both feature extraction and classification was also considered.
The proposed face attribute recognition is also applied in this study for the purpose of detection of the selected attributes that are indicative of the prolonged illicit drug abuse, using the public database FacesOfMeth. Both the deep neural networks for feature extraction and attribute detection, and machine reasoning performed using Bayesian networks are applied. The feasibility and performance of the proposed approach on the public databases of faces with labeled attributes is evaluated in terms of accuracy, precision, sensitivity and specificity.
Description
Keywords
Citation
Tekkam Gnanasekar, S. (2019). Facial Attribute Recognition and its Application in Drug Abuse Detection (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.