Face Detection Research Articles

Robust Face Detection and Identification under Occlusion using MTCNN and RESNET50

In today's rapidly evolving world, where technology is progressing swiftly, there is an increasing demand for facial recognition systems. Technologies are similar to digital forensics in that they can recognize people by scanning faces. However, one key problem they confront is dealing with covered or occluded faces, which might restrict recognition of faces in real-world situations. To overcome this issue, we created a system that is capable of identifying individuals even when their faces are veiled. We used the face detector algorithm called Multi-Task Cascaded Convolutional Neural Network (MTCNN) for face detection with 99.8% accuracy. Further we have conducted feature extraction and pre-processing on our self-created dataset. Our project utilizes the power of deep learning model: Residual Network (ResNet50), the form of deep neural network architectures well-suited for the job of features extraction. These features are matched by using Cosine similarity with accuracy of 92%. By leveraging the capabilities in the deep learning algorithms, this project provides a robust solution for automating the recognition of partially occluded faces.

Designing Face Detection Systems with Gray Wolf Optimization

The main objective of this paper project was to create a state-of-the-art face identification technique that can handle the various difficulties caused by changes in illumination, occlusions, and facial emotions. Face detection is a cornerstone of computer vision, facilitating diverse applications ranging from surveillance systems to human-computer interaction. Throughout this paper, the comprehensive exploration of advancing face detection methodologies has been undertaken, culminating in developing and evaluating a novel approach. The challenges posed by variations in facial expressions, lighting conditions, and occlusions necessitated a multifaceted solution. Our proposed method, which consists of interconnected steps, works quite well to overcome these challenges. Using deep learning architectures to increase feature extraction and discrimination was beneficial in the initial stage of fine-tuning Residual Networks (ResNet-50) to serve as the Region-based Convolutional Neural Network (Faster R-CNN) framework classifier. The process of gradually optimizing thresholds, such as batch size, learning rate, and detection threshold, involved using the Gray Wolf optimization technique (GWO). The conversion process was accelerated and improved overall detection process efficiency and accuracy using a clever fusion of machine learning and metaheuristic optimization techniques. A key component of our methodology is the careful data processing, which was necessary to ensure. The suggested method was carefully examined on a particular dataset, and the 94% training accuracy that was attained together with an identical test dataset accuracy highlights the method's resilience. These findings support the effectiveness of our approach in reducing false positives and negatives, resulting in unmatched recall and precision in the detection system. The discovery has significant significance as it can potentially improve face detection systems' performance and reliability in various real-world applications, such as human-computer interaction and surveillance. Convolutional neural networks, deep learning architectures, and metaheuristic optimization approaches were synergized to produce a new and reliable solution.

A deep learning based detection algorithm for anomalous behavior and anomalous item on buses

This paper proposes a new strategy for analysing and detecting abnormal passenger behavior and abnormal objects on buses. First, a library of abnormal passenger behaviors and objects on buses is established. Then, a new mask detection and abnormal object detection and analysis (MD-AODA) algorithm is proposed. The algorithm is based on the deep learning YOLOv5 (You Only Look Once) algorithm with improvements. For onboard face mask detection, a strategy based on the combination of onboard face detection and target tracking is used. To detect abnormal objects in the vehicle, a geometric scale conversion-based approach for recognizing large-size ab-normal objects is adopted. To apply the algorithm effectively to real bus data, an embedded video analysis system is designed. The system incorporates the proposed method, which results in improved accuracy and timeliness in detecting anomalies compared to existing approaches. The algorithm’s effectiveness and applicability is verified through comprehensive experiments using actual video bus data. The experimental results affirm the validity and practicality of the pro-posed algorithm.

Research on small-scale face detection methods in dense scenes

Research on small-scale face detection methods in dense scenes

LittleFaceNet: A Small-Sized Face Recognition Method Based on RetinaFace and AdaFace

For surveillance video management in university laboratories, issues such as occlusion and low-resolution face capture often arise. Traditional face recognition algorithms are typically static and rely heavily on clear images, resulting in inaccurate recognition for low-resolution, small-sized faces. To address the challenges of occlusion and low-resolution person identification, this paper proposes a new face recognition framework by reconstructing Retinaface-Resnet and combining it with Quality-Adaptive Margin (adaface). Currently, although there are many target detection algorithms, they all require a large amount of data for training. However, datasets for low-resolution face detection are scarce, leading to poor detection performance of the models. This paper aims to solve Retinaface’s weak face recognition capability in low-resolution scenarios and its potential inaccuracies in face bounding box localization when faces are at extreme angles or partially occluded. To this end, Spatial Depth-wise Separable Convolutions are introduced. Retinaface-Resnet is designed for face detection and localization, while adaface is employed to address low-resolution face recognition by using feature norm approximation to estimate image quality and applying an adaptive margin function. Additionally, a multi-object tracking algorithm is used to solve the problem of moving occlusion. Experimental results demonstrate significant improvements, achieving an accuracy of 96.12% on the WiderFace dataset and a recognition accuracy of 84.36% in practical laboratory applications.

Impact of Artificial Intelligence (AI) Image Enhancing Filters on Patient Expectations for Plastic Surgery Outcomes.

Artificial intelligence (AI) technologies use a three-part strategy for facial visual enhancement: (1) Facial Detection, (2) Facial Landmark Detection, and (3) Filter Application (Chen in Arch Fac Plast Surg 21:361-367, 2019). In the context of the surgical patient population, open-source AI algorithms are capable of modifying or simulating images to present potential results of plastic surgery procedures. Our primary aim was to understand whether AI filter use may influence individuals' perceptions and expectations of post-surgical outcomes. We utilized Amazon's Mechanical Turk platform and collected information on prior experience using AI-driven visual enhancement. The cohort was divided into two groups: AI-exposed and non-AI-exposed. Questions gauged confidence in plastic surgery's ability to meet participant expectations. A second survey exposed users to either AI-enhanced or to unenhanced pre-operative photographs. Then, unedited post-operative photographs were shown and surgery's ability to enhance appearance was assessed. A multivariable linear analysis was constructed to measure associations between exposure to AI enhancement and survey outcomes. A total of 426 responses were analysed: 66.9% with AI exposure and 33.1% with no prior exposure. Participants with previous experience using AI-driven enhancers had a significantly higher average score for expectations after plastic surgery (P < 0.001). This finding was true across all outcomes, including surgery's ability to relieve discomfort with appearance/self-esteem (P < 0.001), to avoid post-operative complications (P < 0.001), to decrease post-operative scarring (P < 0.001), and to improve overall appearance (P < 0.001). The image comparison survey revealed that post-operative images were viewed as more successful at improving appearance when no pre-operative filter was applied (P = 0.151). Exposure to AI photograph enhancement may significantly raise expectations for plastic surgery outcomes and may predispose to having lower satisfaction after surgery. The significance of this study lies in its potential to reveal the extent to which AI technologies can shape patient understanding of their plastic surgery outcomes. Plastic surgeons aware of the effect of AI enhancement may consider using these results to guide counselling. his journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .

Local flap guidance for scalp reconstruction following Mohs micrographic surgery.

Keratinocyte carcinomas (KCs) are commonly located on the scalp and often treated with excision with peripheral and deep en face margin assessment (PDEMA), with Mohs micrographic surgery (MMS) being the most frequently used method. Resection of these malignancies results in wounds with a wide variety of sizes, ranging from small, sub-centimeter defects, to extensive, nearly complete scalp defects. MMS is often the preferred treatment for tumor resection and margin clearance, as it allows for maximal healthy tissue preservation and has the lowest recurrence rates. Multiple local flap reconstruction options exist, and the selection depends on the size and location of the defect. This review highlights the indications, typical locations, potential risks, and advantages of frequently used local flaps on the scalp, accompanied by artistic representations of each flap herein. An exhaustive list of the local scalp flaps discussed in this review includes single, bilateral, and triple rotation flaps, the pinwheel flap, the double hatchet flap, the triple rhomboid flap, both the unilateral and bilateral banner flaps with split thickness skin grafts, A to T flaps, H-plasty flaps, single and double bridge flaps, and the pacman flap. Inelasticity of the scalp inhibits tissue movement so flaps on the scalp must be designed to be much larger than flaps in most other facial locations. One of the greatest benefits of working with local flaps on the scalp is often the presence of hair-bearing skin which provides the reconstructive surgeon the ability to hide significant scars under hair-bearing scalp, and the ability to perform wound repair under local anesthetic immediately following tumor removal by MMS.

Robust Face Tracking Under Challenging Conditions Using Linear Regression and YOLO algorithm

Face detection and tracking play a crucial role in various computer vision applications, including surveillance, fault face detection systems, artificial intelligence, etc. The objective of this paper is to enhance the precision of face detection and tracking through the introduction of an innovative approach centered on the linear regression algorithm. The effectiveness of the proposed method was compared to the traditional Kalman filter approach. Additionally, the study explored the integration of the YOLO algorithm for face detection with the linear regression tracking algorithm to further enhance accuracy. The proposed algorithm's performance is assessed through comprehensive experiments on annotated images and video sequences affected by occlusions or other issues such as poor lighting conditions and motion blur. These experiments utilize the COCO dataset, operating at a speed of 60 FPS. The experimental results show that the proposed method can accurately track the human face in different facial positions

Face Detection and Localization in Video Using HOG with CNN

Face Detection and Localization in Video Using HOG with CNN

The early development of a combined micro- and full-field X-ray fluorescence analysis system using white X-rays at PLS-II.

X-ray fluorescence (XRF) is widely used to analyze elemental distributions in samples. Micro-XRF (µ-XRF), the most basic conventional XRF technique, offers good spatial resolution through precise 2D scanning with a micrometre-sized X-ray source. Recently, synchrotron based XRF analysis platforms have achieved nano-XRF with highly focused X-rays using polycapillary optics or mirrors, leveraging the excellent coherence of synchrotron radiation. However, XRF techniques are hindered by long data acquisition times (exceeding several hours) due to their point-by-point scanning approach, impeding large-area elemental mapping. Full-field XRF (FF-XRF), developed in the 2010s and based on the high brilliance of synchrotron X-rays, enables significantly shorter (less than a few minutes) data acquisition times via single-exposure imaging using a 2D X-ray detector. Nevertheless, it is constrained by relatively low spatial resolution and sensitivity. Hence, a new XRF platform is required to accommodate resolution demands to cover diverse experimental purposes. In this study, we developed a preliminary model of a novel XRF system that combines micro- and full-field XRF setups to address these limitations. This system allows easy mode switching while maintaining the region of interest of the imaging system within a single apparatus, simply by rotating the sample to face either detector depending on research purposes. We anticipate that this new XRF system will be widely utilized in various research fields as the initial XRF setup at Pohang Light Source-II.

Raspberry Pi-based Facial Recognition for Vehicle Access Control

The integration of advanced biometric security systems in modern vehicles has become crucial to enhance safety and user convenience. This paper presents the design and implementation of a Raspberry Pi-based facial recognition system for vehicle access control, utilizing the power of computer vision and machine learning to provide secure, keyless vehicle entry. The system employs a Raspberry Pi microcontroller, coupled with a camera module, to capture and analyze the driver's facial features. OpenCV, a widely-used computer vision library, is integrated with a pre-trained deep learning model to perform facial detection and recognition in real-time. The system compares the captured facial data with pre-registered user profiles stored in a secure database. Upon a successful match, the Raspberry Pi sends a signal to unlock the vehicle, ensuring seamless and secure access. The necessity of adopting facial recognition technology for vehicle access lies in its non-invasive, hands-free nature, providing a more convenient and secure alternative to traditional key-based systems. Moreover, it adds an additional layer of security, minimizing the risk of unauthorized access through stolen keys or fob duplication. Experimental results demonstrate that the facial recognition system achieves a high accuracy rate of 95% under various lighting conditions and user angles. The average processing time for facial recognition is less than 1 second, ensuring swift vehicle access. Additionally, the system exhibits robust performance even with multiple registered users, confirming its scalability and practicality for real-world applications. This research highlights the potential of Raspberry Pi as a cost-effective, efficient platform for implementing facial recognition systems in vehicles, paving the way for further innovations in automotive security. Key Words: Raspberry Pi, Webcam, Image processing, the Main module

FROM ORIGINS TO INNOVATIONS: AI'S ROLE AND THE COST IMPACT ON COMPUTER VISION

The field of Computer Vision, a pivotal subdomain of Artificial Intelligence (AI), has seen extraordinary advancements since its emergence in the 1960s. This paper examines the historical development of Computer Vision technologies, tracing the journey from early foundational models, such as Frank Rosenblatt’s Perceptron, to contemporary breakthroughs driven by Deep Learning. Key milestones are explored, including the development of algorithms like Scale-Invariant Feature Transform (SIFT), Viola-Jones for face detection, and Eigenfaces, which paved the way for modern solutions such as Convolutional Neural Networks (CNNs), YOLO and FaceNet. The paper highlights the evolution of face detection and recognition techniques, contrasting traditional methods with the transformative capabilities of Deep Learning-driven approaches. Additionally, we analyze the growing computational demands of modern algorithms, discussing the trade-offs between accuracy and efficiency and their implications for practical applications. This study underscores the rapid progression of Computer Vision, its challenges, and its role as a cornerstone in shaping the future of Artificial Intelligence.

Optimizing Face Recognition with PCA and KNN: A Machine Learning Approach

Face detection and recognition have become critical applications in various fields, including security, identity verification, and human-computer interaction. This paper presents a comprehensive analysis of face detection techniques using Artificial Intelligence (AI), focusing on the integration of PCA and KNN algorithms. PCA is employed to reduce the dimensionality of face image datasets, effectively extracting important features while minimizing data loss. The KNN classifier is used for classification by identifying the closest matching face in a dataset. By applying these techniques to the LFW dataset, we achieved an overall accuracy of 88%, demonstrating the efficacy of this approach for face detection. The methodology involves training the system with face image data, utilizing PCA to project the images onto a lower-dimensional space, and applying KNN to classify the images based on their reduced feature set. The implementation was carried out using Python’s Scikit-learn library, highlighting the ease of combining well-established machine learning algorithms in a straightforward programming environment. Results show that using KNN with an optimal K value of 5, alongside PCA retaining 95% variance, provides a robust and efficient solution for face detection tasks. While this approach achieved significant success, further improvements could be made by integrating advanced classifiers such as CNNs or exploring neural networks for feature extraction. Additionally, real-time performance can be enhanced by optimizing the computational process or leveraging OpenCV for real-world applications.

DETECTION OF AI GENERATED IMAGES USING DEEP LEARNING

This Gradio app detects fake images using a fine-tuned InceptionResnetV1 model with VGGFace2 weights. It employs MTCNN for face detection and extraction, resizing the face for model compatibility. Grad-CAM provides explainability by highlighting face areas influencing predictions, overlaying a heatmap on the original image. The model outputs confidence scores for "real" or "fake" classifications. Results and visualizations are displayed in an interactive Gradio interface. Keywords- Deep fake Detection ,Convolutional Neural Networks (CNNs),Generative Adversarial Networks(GANs), Image Forensics, Anomaly Detection, Artifact Detection, Feature Extraction, Visual Inconsistencies, Adversarial Robustness.

Lightweight Research on Fatigue Driving Face Detection Based on YOLOv8

Introduction: With the rapid development of society, motor vehicles have become one of the main means of transportation. However, as the number of motor vehicles continues to increase, traffic safety accidents also continue to appear, bringing serious threats to people's lives, and property safety. Fatigue driving is one of the important causes of traffic safety accidents. Method: To address this problem, a target detection algorithm called VA-YOLO is designed to improve the speed and accuracy of facial recognition for fatigue checking. The algorithm employs a lightweight backbone network, VanillaNet, instead of the traditional backbone network, which reduces the computational and parametric quantities of the model. The SE attention mechanism is also introduced to enhance the model's attention to the target features, which further improves the accuracy of target detection. Finally, in terms of the bounding box regression loss function, the SIoU loss function is used to reduce the error. Result: The experimental results show that, compared toYolov8n, the VA-YOLO algorithm improves the accuracy by 1.3% while the number of parameters decreases by 30%. Conclusion: This shows that the VA-YOLO algorithm has a significant advantage in realizing the balance between the number of parameters and accuracy, which is important for improving the speed and accuracy of fatigue driving detection.

Pengembangan Sistem Ujian Online Multibahasa dengan Deteksi Wajah di Universitas Nurul Jadid

The digitalization of academic systems enhances efficiency and reliability, particularly in language competency testing. This study develops an advanced multilingual online exam system integrated with face and voice recognition technologies at Universitas Nurul Jadid. Employing the Waterfall model, the system was designed, programmed, tested, and maintained to address challenges such as network instability and delayed scoring. The system supports English, Arabic, and Mandarin language exams and includes automatic certificate generation. Developed using Laravel 11 and TensorFlow for face detection, the system improves the reliability and security of online exams while maintaining functionality under low network conditions. Testing revealed significant improvements in efficiency and accuracy, although enhancements to user interface responsiveness and scalability for handling high test-taker volumes remain necessary. This research provides a scalable framework for other institutions aiming to modernize academic evaluations.

Task-aware image quality estimators for face detection

Understanding the quality of face images plays a critical role in enhancing the efficiency of end-to-end face analytics systems, which perform tasks such as face detection, alignment, and recognition in a sequential manner. Recently, the development of face quality estimators (QEs) specifically for face recognition has received significant attention. However, in end-to-end face analytics systems, the performance of the face detector directly affects the face recognizer. Thus, assessing the suitability of an image for face detection before passing it on to face recognition can improve the resource utilization in such systems. In this research, we first introduce the detectability (DET) score of an image, a novel quality metric that links image quality to face detection performance. We use this DET score to design two novel QEs for face detection: supervised face detection quality estimator (SFDQE) and unsupervised face detection quality estimator (UFDQE). We also propose the mAP vs. reject protocol (mvR), a systematic evaluation protocol for assessing QEs in the context of face detection. In our experiments, we illustrate the effectiveness of SFDQE and UFDQE in determining the suitability of an image for face detection. Furthermore, we show the ability of our QEs to generalize; each is a powerful tool for image quality estimation in general object detection scenarios.

Differential late-stage face processing in autism: a magnetoencephalographic study of fusiform gyrus activation

BackgroundAutism is associated with alterations of social communication, such as during face-to-face interactions. This study aimed to probe face processing in autistics with normal IQ utilizing magnetoencephalography to examine event-related fields within the fusiform gyrus during face perception.MethodsA case–control cohort of 22 individuals diagnosed with autism and 20 age-matched controls (all male, age 29.3 ± 6.9 years) underwent magnetoencephalographic scanning during an active task while observing neutral faces, face-like pareidolic objects, and non-face objects. The fusiform face area was identified using a face localizer for each participant, and the cortical activation pattern was normalized onto an average brain for subsequent analysis.ResultsEarly post-stimulus activation amplitudes (before 100–200 ms) indicated differentiation between stimuli containing fundamental facial features and non-face objects in both groups. In contrast, later activation (400–550 ms) differentiated real faces from both pareidolic and non-face objects across both groups and faces from objects in controls but not in autistics. There was no effect of autistic-like traits.ConclusionsThe absence of group differences in early activation suggest intact face detection in autistics possessing a normal IQ. Later activation captures a greater degree of the complexity and social information from actual faces. Although both groups distinguished faces from pareidolic and non-face objects, the control group exhibited a slightly heightened differentiation at this latency, indicating a potential disadvantage for autistics in real face processing. The subtle difference in late-stage face processing observed in autistic individuals may reflect specific cognitive mechanisms related to face perception in autism.

Facial Detection and Recognition Analysis using Ontology-Driven Machine Learning Model

Facial detection and recognition technologies have achieved remarkable advancements through the integration of ontology-driven machine learning (ML). This study examines how ontology structured framework for representing domain-specific knowledge to enhance the robustness, accuracy, and interpretability of ML models, particularly Convolutional Neural Networks (CNNs), in the field of facial detection and recognition. Acknowledging the limitations of traditional ML methods, such as data inefficiency, inadequate generalization, and insufficient interpretability, this research addresses critical challenges, including semantic discrepancies, variations in environmental conditions, and ethical considerations. By leveraging ontology, the study aims to provide semantic enrichment, contextual adaptation, and data augmentation for ML models. A hybrid methodology is introduced that effectively integrates ontology with CNNs, in conjunction with the Viola-Jones and Eigenfaces algorithms, to improve performance in facial recognition tasks. The study utilizes comprehensive datasets such as CelebA and MegaFace, employing rigorous preprocessing and training processes that incorporate ontology-based feedback mechanisms. The results demonstrated significant improvements in accuracy, robustness, and explainability. Ontology-CNN models outperform traditional approaches in managing variations in facial attributes and environmental conditions. This study concludes that ontology-based frameworks present a promising avenue for developing efficient and ethically responsible facial recognition systems. Future research should focus on exploring the scalability of these models across diverse demographic contexts and further addressing ethical considerations related to privacy and fairness.

Account Login and Database Access Control System with Time Attendance Through Facial Recognition

The traditional method of recording attendance faces challenges, such as errors, lack of accuracy, and time consumption, leading to low efficiency and obsolescence. To address these issues, this study proposes the development of an automated attendance system that uses real-time face detection technology. The goal is to enhance efficiency, accuracy, and security by incorporating automated facial detection for attendance tracking. The attendance system is based on Haar Cascades face detection classifier technology for facial recognition. Attendance data is stored in a MySQL database that is managed through SQL programming. By detecting faces and updating attendance records in a timely manner, users can also access their pages. Face detection entails recognizing faces through real-time monitoring. When a face is detected, users are prompted to enter their names to register attendance as the system updates the MySQL database. Accuracy tests conducted under different conditions showed an accuracy rate of up to 97.44%. Attendance details are stored in a database containing the names of attendees, along with date and time stamps of their attendance. For improvement, it is proposed in the future to include security measures such as anti-impersonation and mask detection techniques along with incorporating deep learning principles. This will help enhance the security and effectiveness of the systems in situations.