Gender and ethnicity classification using deep learning in heterogeneous face recognition

Narang, Neeru; Bourlai, Thirimachos

doi:10.1109/icb.2016.7550082

Cited by 44 publications

(22 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This face recognition sensitivity has induced serious care from the biometric society with several publications on countermeasure studies being published [6]. The issue of soft biometric traits classification related to age, ethnicity, gender, and odour has not yet been a resolved due to problematic exposure conditions [15,21]. The biometric benefit considered more stable and stronger enough not to change or lose.…”

Section: Effects Of Handling Missing Values Of Vocs Gases Emitted Fromentioning

confidence: 99%

Effects of Handling Missing Values of VOCS Gases Emitted From Human for Human Detection

Sabri¹,

Alfred²

2019

IJRTE

View full text Add to dashboard Cite

The main objective of this paper is to investigate the effects of replacing missing values of Volatile Organic Compounds (VOC) that represent gases emitted from human on the accuracy of classifying individual person. These missing values will be replaced with three possibilities, which include 0, 1, and Random number between 0 and 1. The effects of using these predefined values on the classification accuracy are investigated by conducting experiments that involve applying a list of classification methods to classify 15 humans using human odour. Each person is characterized by their own pre-selected 15 gases emitted from their sweats. In this paper, we also study and determine the minimum number of gases that is required to produce acceptable results to correctly classify an individual person based on the gases emitted from their bodies. Based on the results obtained from the conducted experiments, the maximum and minimum allowable numbers of missing gases in human odour samples in reference to human emitted gases are 4 and 3. The best accuracy result when missing values are introduced in the odour dataset is the ensemble Bagged Trees.

show abstract

Section: Effects Of Handling Missing Values Of Vocs Gases Emitted Fromentioning

confidence: 99%

Effects of Handling Missing Values of VOCS Gases Emitted From Human for Human Detection

Sabri¹,

Alfred²

2019

IJRTE

View full text Add to dashboard Cite

show abstract

“…This face recognition vulnerability has induced serious attention from the biometric community with several papers on countermeasure studies being published [4]. The issue of soft biometric traits classification related to age, ethnicity, gender, and odour has not yet been a resolved due to problematic exposure conditions [13,19]. The biometric benefit is that it doesn't lose or change.…”

mentioning

confidence: 99%

Odour Based Human Identification and Classification using Neural Networks

2019

IJRTE

View full text Add to dashboard Cite

Biometrics permits an individual to be authenticated and identified by computer systems following on a set of verifiable and identifiable data that are precise and unique in nature. This mechanism constitutes a cutting-edge method of identifying an individual since it precisely establishes more explicit and direct connection with humans than mere passwords since biometrics tend to use measurable behavioral and physiological characteristics of human. In this paper, a framework for human identification is proposed distinctively based on specific human odour features. 15 samples of female and male human odour are collected from different age groups, only 15 effective Volatile Organic Compounds (VOCs) are chosen. In this paper, several diverse functions of neural network activation are tested such as Levenberg-Marquardt backpropagation, Gradient descent backpropagation, and Resilient backpropagation. Besides, numerous neural network topologies are tested by means of variety hidden layers and different number of neurons and. Different energy functions were tested TAN- Sigmoid transfer, Linear transfer, and LOG- Sigmoid transfer. Considering the obtained results, employing two hidden layers with more neurons in the hidden layers- to be specific: 15 neurons in every layer- has yielded better accuracy in performance with an accuracy rate of 100%. The unsurpassed framework for algorithm learning to be used for human identification can be backpropagation learning algorithm named the Levenberg-Marquardt. The best function for activation established in this paper is the function of TANSigmoid transfer. The performance accuracy consistency in recognizing human can be enhanced using a big number of study samples.

show abstract

“…Developed a scenario dependent and sensor adaptable convolutional neural network (CNN) to perform the data filtering. While the manual filtering can be done, such a process is time consuming, especially when dealing with large datasets [22,23]. Therefore an automated way of classifying images into specific scenario is needed.…”

Section: Study 2: Developed Deep Features Based Data Filtering Networkmentioning

confidence: 99%

“…The live subject-capture setup using the VIS and mid-range NIR cameras. At the bottom can see a set of face image samples acquired by our system at a night time environment and at variable standoff distances[69].…”

mentioning

confidence: 99%

“…To address this problem, a deep learning based method for the demographic filtering is developed. The focus of this chapter is to discuss implementation of the developed convolutional neural network for demographic filtering from Study 2 and to determine whether the usage of demographic filtering can enhance recognition performance[69]. The architecture for this study is represented inFig.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Erbium alloyed AlN thin films: Structural, piezoelectric and magnetic properties

Narang¹

View full text Add to dashboard Cite

Deep Features based Hierarchical Classification Scheme for Face Recognition in Heterogeneous Environments Neeru Narang This dissertation investigates the advantages and limitations of the heterogeneous problem of matching infrared long-range, night time face images against their visible counterparts. The contributions of the thesis are threefold. First, a multi-feature scenario dependent fusion scheme is developed, where Gabor Wavelets, Histogram of gradients (HOG) and Local binary patterns (LBP) feature descriptors are empirically selected. Next, a set of fusion score level schemes are developed and applied before face matching. The developed fusion scheme results in 54 percent point (pp) higher recognition performance than the baseline established using a commercial and a set of academic face matchers. Second, to further improve the performance of baseline face recognition (FR) systems, a scenario dependent and sensor adaptable convolutional neural network (CNN) is developed that groups the data in terms of demographic information, including scenarios (situations) such as indoors or outdoors data, as well as distance and sensor based data. The automated grouping scheme developed is applied before the FR algorithms are used, improving baseline performance from 48% (all vs. all) to 70% (with data grouping). Third, an image quality restoration scheme is designed and developed. This scheme is beneficial to FR systems because the quality of face data captured under challenging conditions is affected by a variety of noise factors (including low illumination conditions, variable standoff distances, and uncooperative subjects). Thus, image quality is responsible for the performance degradation of conventional FR matchers. The developed scheme improves, first, the quality of distorted face images and, then, FR performance in terms of the rank-1 identification rate. Based on the experimental results the major conclusion from this research is that the aforementioned schemes discussed in this dissertation significantly improve cross-spectral face matching performance on diverse scenarios, when used either independently or in combination. The experimental results are further supported by statistical analysis tests, conducted to find the statistical significance of incorporating the developed score level fusion schemes, as well as the demographic filtering to FR systems.

show abstract

Gender and ethnicity classification using deep learning in heterogeneous face recognition

Cited by 44 publications

References 10 publications

Effects of Handling Missing Values of VOCS Gases Emitted From Human for Human Detection

Effects of Handling Missing Values of VOCS Gases Emitted From Human for Human Detection

Odour Based Human Identification and Classification using Neural Networks

Erbium alloyed AlN thin films: Structural, piezoelectric and magnetic properties

Contact Info

Product

Resources

About