Development of Deep Learning Algorithms for Improved Facial Recognition in Security Applications

Emotion recognition through facial expression analysis is an emerging research in Artificial Intelligence which faces many challenges. The problem is the variation of facial expressions that displays human emotions. Humans can subjectively express the same emotions in various ways. To overcome the problem of ambiguity in emotion expression, a fuzzy approach is developed to analyze the facial components in determining the type of emotion. In this study, we proposed a framework for fuzzy emotion recognition as a representation of the psychologist knowledge. Three stages in the fuzzy emotion recognition were facial feature extraction with Active Appearance Model; Semantic-linguistic facial features extraction; fuzzy emotion recognition with Fuzzy Emotion Classification. System performance testing provided the best results on extended Cohn Kanade (CK+) facial expression dataset, with the accuracy of linguistic facial component recognition 0.98, and accuracy of fuzzy emotion recognition 0.90. Testing was also performed on custom-made Indonesian Mixed Emotion Dataset (IMED) which resulted in accuracy of 0.87. The fuzzy emotion recognition has a potential to be applied in various real problems such as virtual counseling, stress detection, lie detection, and e-commerce.

Facial recognition technology has advanced significantly due to the development of deep learning algorithms. This paper explores deep learning, a branch of machine learning that employs multi-layered neural networks to simulate human decision-making processes in facial recognition. It provides a brief literature review of significant works in various deep learning architectures, including Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks. The core of the study is the implementation of the GhostFaceNets model, an enhancement of GhostNets, which is specifically designed for efficient and accurate facial recognition. By using Ghost Modules, this model reduces computational redundancy in generating additional feature maps through linear operations. An integrated attention mechanism is used in this study to emphasize critical facial features. Additionally, this study also employs the ArcFace loss function to improve class separation accuracy. The VGG2-FP dataset was used to train and evaluate this model and achieved an accuracy of 94.45%. This study contributes to the evolution of facial recognition systems, particularly in constrained computational environments.

Face recognition (FR) is a method that uses face feature analysis and comparison to identify or verify individuals. Siamese neural networks (SNNs) are an effective method for FR, providing high accuracy and versatility, especially in situations where data is restricted. Unlike standard neural networks, SNNs learn to distinguish between pairs of inputs rather than individual inputs. However, detecting and recognizing faces in unconstrained environments poses a significant challenge due to various factors such as head pose, illumination, and facial expression variations. The aim of this paper is to design and develop an efficient approach based on SNNs and Transfer Learning methods. For this purpose LFW dataset and transfer learning architectures like VGG-16, EfficientNet, RestNet50 and ConvNext have been utilised. Performance of the architectures were measured using 5-Fold cross validation. According to results, EfficientNet, RestNet50 and ConvNext produced 78% accuracy, 95% and 93 % accuracy respectively. SNN with VGG-16 exhibited a low loss and produced the best accuracy in face recognition with 96%.

The aim of this study was to determine how the depressor supercilii (DS) connects to the levator labii superioris alaeque nasi (LLSAN) and inferior fibers of the orbicularis oculi (OOc INF) in the human midface. While grimacing, contraction of the DS with fibers connecting to the LLSAN and OOc INF can assist in pulling the medial eyebrow downward more than when these connecting fibers are not present. Contraction of these distinct connecting fibers between the DS and the LLSAN can also slightly elevate the nasal ala and upper lip. The DS was examined in 44 specimens of embalmed adult Korean cadavers. We found that the DS connected to the LLSAN or the OOc INF by muscle fibers or thin aponeuroses in 33 (75.0%) of the 44 specimens. The DS was connected to both the LLSAN and OOc INF by muscle fibers or aponeuroses and had no connection to either in 5 (11.4%) and 11 (25.0%) specimens, respectively. The DS was connected to the LLSAN by the muscle fibers and thin aponeuroses in 6 (13.6%) and 4 (9.1%) specimens, respectively. The DS was connected to the OOc INF by the muscle fibers and thin aponeuroses in 5 (11.4%) and 23 (52.3%) specimens, respectively. Our findings regarding the anatomical connections of the glabellar region DS to the midface LLSAN and OOc INF provide insights on the dynamic balance between the brow depressors such as the DS and brow-elevating muscle and contribute to understanding the anatomical origins of individual variation in facial expressions. These results can also improve the safety, predictability, and aesthetics of treatments for the glabellar region with botulinum toxin type A and can be helpful when performing electromyography.

Compared to more conventional techniques such as fingerprint recognition, the use of biometric authentication systems in computer-based examinations offers several advantages. To overcome the disadvantages of fingerprint biometrics, such as high administrative costs, long authentication times, and accuracy issues, this paper proposes an innovative solution that leverages state-of-the- art facial recognition technology. Although biometric fingerprint systems are reliable to a certain extent, they present significant challenges in the context of computer-assisted examinations. The authentication process can be time-consuming and can result in delays and logistical problems. These systems can also be prone to errors, including false positives and false negatives, compromising the integrity of the investigation process. These limitations require research into more efficient and accurate biometric solutions. The developed system uses FaceNet and Multi- Task Cascaded Convolutional Neural Network (MTCNN) algorithms in achieving superior accuracy, efficiency, as well as security in verifying candidates' identities during exams. FaceNet excels at facial recognition by mapping faces in a compact Euclidean space, ensuring high accuracy even with slight variations in facial expressions, angles or lighting conditions. MTCNN increases the robustness of the system through precise face detection and alignment, which are critical for reliable performance. The results show that the facial recognition system outperforms traditional fingerprint-based methods. Accuracy is significantly improved, reducing misidentifications, while streamlining the authentication process, minimizing delays, and improving overall efficiency. The robustness of the system ensures consistent performance despite environmental fluctuations.

2D Face recognition has been extensively studied for decades and has reached remarkable results in recent years. However, 2D Face recognition is sensitive to variations in poses, facial expressions and illuminations. Depth images provide valuable information to help model facial boundaries and understand the global facial layout and provide low frequency patterns. Intuitively, RGB-D images are more robust to external environments than RGB images. Unfortunately, RGB-D datasets are orders of magnitude smaller than 2D datasets and insufficient to train a deep CNN model as effective as RGB-based models. To tackle these challenges, we present an RGB-D ResNet50 model which can be transferred from a pretrained RGB model and takes RGB-D images as input. We achieved an accuracy of 94.64% and won the \(1^{\text {st}}\) place on 3D Face Recognition Algorithm Challenge, 2019 BenchCouncil International Artificial Intelligence System Challenges.

Facial recognition technologies are increasingly employed in security and law enforcement due to advances in computer vision and machine learning. Despite these advancements, balancing recognition accuracy, computational efficiency, and data security remains a significant challenge in research. This paper proposes a secure and efficient framework for criminal identification that integrates facial keypoint detection, normalization, Principal Component Analysis (PCA), and cryptographic hashing. The approach begins with grayscale conversion and image size standardization, followed by the detection of facial keypoints using pre-trained models to localize discriminative facial features. Normalization ensures scale and pose invariance, while PCA reduces feature dimensionality, improving computational performance without compromising representational fidelity. The reduced feature vectors are subsequently transformed into cryptographic hash values, providing unique, tamper-resistant identifiers for individuals. Experimental results demonstrate that the proposed method achieves high identification accuracy and computational efficiency while maintaining robustness against variations in lighting conditions and facial expressions. In addition, we introduce a hybrid deep feature–PCA fusion stage, where compact embeddings from a lightweight CNN are combined with normalized keypoints before being subjected to secure hashing. This fusion enhances robustness against occlusions and adversarial variations, providing a modern and secure update to traditional biometric pipelines.

The human face has a great accumulation and a diversity of facial expressions. It explores the feelings of a person and can be used to judge the emotional intents of the person to a certain level. By using facial detection and recognition systems, varieties of applications are working in computer vision, surveillance system, security, authentication, or verification of a person and home automation system based on digital image processing with the help of the Internet of Things. The state of the art in these applications is to detect expressions with their intensity level. It is an attention-grabbing problem due to the complex nature of facial features, which is associated with emotions. For that purpose, it is essential to develop an innovative deep learning model to detect and estimate the facial expression intensity level. To do this, a multi-level deep convolutional neural network is proposed to recognize facial expression and their intensity level. At the first level, Expression-Net classifies face expressions, and at the second level, Intensity-Net estimates the intensity of the facial expression. Evaluation of the proposed model for facial expression recognition and intensity estimation is carried out by using the extended Cohn–Kanade and Japanese Female Facial Expression datasets. The proposed method shows an outstanding performance in terms of accuracy of 98.8% and 97.7% for both the datasets as compared to state-of-the-art techniques.

Understanding how customers feel is one of the most valuable insights a business can gain. Traditionally, companies depend on surveys, ratings, or written feedbacks to collect reviews—but these methods are often time-consuming, biased, or ignored by users. With the rapid advancement of Artificial Intelligence (AI) and Machine Learning (ML), it is now possible to automatically detect and interpret human emotions using technologies like facial recognition, video analysis, and Natural Language Processing (NLP). This paper presents a comprehensive survey of the latest research efforts aimed at improving customer feedback systems using AI-powered emotion recognition. Many researchers try to understand emotions such as happiness, sad, anger, neutral, surprise by studying people’s facial expressions, often using images or videos taken from cameras or webcams. Popular techniques include Convolutional Neural Networks (CNNs) for image-based emotion classification, Support Vector Machines (SVMs) for quick categorization. This document gives formatting instructions for authors preparing papers for publication in the Proceedings of an IEEE conference. The authors must follow the instructions given in the document for the papers to be published. You can use this document as both an instruction set and as a template into which you can type your own text. Keywords: Emotion Recognition, Facial Expression Recognition, Machine Learning, CNN, Viola-Jones Algorithm

Facial expression recognition is an effective method for identifying someone's emotional expression. Emotional expressions can be recognized from changes in facial expressions, wrinkles on the forehead, blinking of the eyes, or changes in facial skin color. Facial expressions that a person generally has, such as neutral, angry, happy expressions. The problem that often occurs is the subjective assessment of a person's expression. This research examines how artificial intelligence can recognize facial expressions. The facial recognition process in the research uses a Convolutional Neural Network (CNN), which is a deep learning method capable of carrying out an independent learning process for object recognition, object extraction and classification and can be applied to high resolution images that have a nonparametric distribution model. The two main stages in CNN are feature learning and classification. The results of facial expression recognition can be used to detect a person's emotions. This research uses the FER2013 dataset which contains images of happy, sad, angry, afraid, surprised, disgusted and neutral emotions. The data set in the research received tests that had been carried out, namely the percentage of accuracy level in the model was 76%. It is hoped that the classification of emotions resulting from this research can contribute to the development of artificial intelligence technology and as a tool in various fields such as psychology, education and others. For further research, it can be developed further by adding other architectures such as VGG19, MobileNet, and ResNet-50 so that the resulting CNN model is more optimal.

Face recognition is a kind of biometric technology that recognizes identities through human faces. At first, the speed of machine recognition of human faces was slow and the accuracy was lower than manual recognition. With the rapid development of deep learning and the application of Convolutional Neural Network (CNN) in the field of face recognition, the accuracy of face recognition has greatly improved. FaceNet is a deep learning framework commonly used in face recognition in recent years. FaceNet uses the deep learning model GoogLeNet, which has a high accuracy in face recognition. However, its network structure is too large, which causes the FaceNet to run at a low speed. Therefore, to improve the running speed without affecting the recognition accuracy of FaceNet, this paper proposes a lightweight FaceNet model based on MobileNet. This article mainly does the following works: Based on the analysis of the low running speed of FaceNet and the principle of MobileNet, a lightweight FaceNet model based on MobileNet is proposed. The model would reduce the overall calculation of the network by using deep separable convolutions. In this paper, the model is trained on the CASIA-WebFace and VGGFace2 datasets, and tested on the LFW dataset. Experimental results show that the model reduces the network parameters to a large extent while ensuring the accuracy and hence an increase in system computing speed. The model can also perform face recognition on a specific person in the video.

An efficient face recognition system should recognize faces in different views and poses. The efficiency of a human face recognition system depends on the capability of face recognition in presence of changes in the appearance of face due to expression, pose and illumination. A novel algorithm which utilizes the combination of texture and depth information based on Modular PCA to overcome the problem of pose variation and illumination change for face recognition is proposed. The system has combined 2D and 3D systems in the feature level which presents higher performance in contrast with methods which utilizes either 2D or 3D system separately. A multimodal face recognition based on Modular PCA when compared with conventional PCA algorithm has an improved recognition rate for face images with large variations in illumination and facial expression. The proposed algorithm is tested with FRAV3D database that has faces with pose variation and illumination changes. Recognition rates from experimental results show the superiority of Modular PCA over conventional PCA methods in tackling face images with different pose variations and changes in illuminations. The proposed algorithm shows a recognition rate of 86% that is achieved in fusion experiment.

Deep learning, in particular Convolutional Neural Network (CNN), has achieved promising results in face recognition recently. However, it remains an open question: why CNNs work well and how to design a 'good' architecture. The existing works tend to focus on reporting CNN architectures that work well for face recognition rather than investigate the reason. In this work, we conduct an extensive evaluation of CNN-based face recognition systems (CNN-FRS) on a common ground to make our work easily reproducible. Specifically, we use public database LFW (Labeled Faces in the Wild) to train CNNs, unlike most existing CNNs trained on private databases. We propose three CNN architectures which are the first reported architectures trained using LFW data. This paper quantitatively compares the architectures of CNNs and evaluate the effect of different implementation choices. We identify several useful properties of CNN-FRS. For instance, the dimensionality of the learned features can be significantly reduced without adverse effect on face recognition accuracy. In addition, traditional metric learning method exploiting CNN-learned features is evaluated. Experiments show two crucial factors to good CNN-FRS performance are the fusion of multiple CNNs and metric learning. To make our work reproducible, source code and models will be made publicly available.

Face recognition has many important applications in areas such as public surveillance and security, identity verification in the digital world, and modeling techniques in multimedia data management. Facial expression recognition is also important for targeted marketing, medical analysis, and human–robot interaction. In this paper, we survey a few techniques for facial analysis. We compare the cloud platform AWS Rekognition, convolutional neural networks, transfer learning from pre-trained neural nets, and traditional feature extraction using facial landmarks for this analysis. Although not comprehensive, this survey covers a lot of ground in the state-of-the-art solutions for facial analysis. We show that to get high accuracy, good-quality data and processing power must be provided in large quantities. We present the results of our experiments which have been conducted over six different public as well as proprietary image data sets.

One of the ways humans communicate is by using facial expressions. Research on technology development in artificial intelligence uses deep learning methods in human and computer interactions as an effective system application process. One example, if someone does show and tries to recognize facial expressions when communicating. The prediction of the expression or emotion of some people who see it sometimes does not understand. In psychology, the detection of emotions or facial expressions requires analysis and assessment of decisions in predicting a person's emotions or group of people in communicating. This research proposes the design of a system that can predict and recognize the classification of facial emotions based on feature extraction using the Convolution Neural Network (CNN) algorithm in real-time with the OpenCV library, namely: TensorFlow and Keras. The research design implemented in the Raspberry Pi consists of three main processes, namely: face detection, facial feature extraction, and facial emotion classification. The prediction results of facial expressions in research with the Convolutional Neural Network (CNN) method using Facial Emotion Recognition (FER-2013) were 65.97% (sixty-five point ninety-seven percent).