The recent success of convolutional neural networks has led to the development a variety new effective and efficient architectures. However, few them have been designed for specific case face recognition. Inspired on state-of-the-art ShuffleNetV2 model, lightweight architecture is presented in this paper. proposal, named ShuffleFaceNet, introduces significant modifications order improve recognition accuracy. First, Global Average Pooling layer replaced by Depth-wise Convolution layer,...

Fingerprint-based biometric systems have experienced a large development in the past. In spite of many advantages, they are still vulnerable to attack presentations (APs). Therefore, task determining whether sample stems from live subject (i.e., bona fide) or an artificial replica is mandatory requirement which has recently received considerable attention. Nowadays, when materials for fabrication Presentation Attack Instruments (PAIs) been used train Detection (PAD) methods, PAIs can be...

Precise instrument segmentation aids surgeons to navigate the body more easily and increases patient safety. While accurate tracking of surgical instruments in real-time plays a crucial role minimally invasive computer-assisted surgeries, it is challenging task achieve, mainly due to: (1) complex environment, (2) model design trade-off terms both optimal accuracy speed. Deep learning gives us opportunity learn environment from large surgery scene environments placements these real world...

Given the current COVID-19 pandemic, most people wear a mask to effectively prevent spread of contagious disease. This sanitary measure has caused significant drop in effectiveness face recognition methods when handling masked faces on practical applications such as access control, attendance, and authentication-based mobile payment. Under this situation, recent efforts have been focused boosting performance existing technology faces. Some solutions trying tackle issue fine-tune deep...

Artificial intelligence’s rapid advancement has enabled various applications, including intelligent video surveillance systems, assisted living, and human–computer interaction. These applications often require one core task: video-based human action recognition. Research in recognition is vast ongoing, making it difficult to assess the full scope of available methods current trends. This survey concisely explores vision-based field defines concepts, definitions explanations common challenges...

Image-based tracking of laparoscopic instruments plays a fundamental role in computer and robotic-assisted surgeries by aiding surgeons increasing patient safety. Computer vision contests, such as the Robust Medical Instrument Segmentation (ROBUST-MIS) Challenge, seek to encourage development robust models for purposes, providing large, diverse, annotated datasets. To date, most existing instance segmentation medical were based on two-stage detectors, which provide results but are nowhere...

In the past decade, research in face recognition area has advanced tremendously, particularly uncontrolled scenarios (face wild). This advancement been achieved partly due to massive popularity and effectiveness of deep convolutional neural networks availability larger unconstrained datasets. However, several challenges remain context very low resolution homogeneous (same domain) heterogeneous (different recognition. this survey, we study seminal novel methods tackle problem provide an...

Presentation Attack Detection (PAD) is the task of determining whether a sample stems from live subject (bona fide presentation) or an artificial replica (Presentation Instrument, PAI). Several PAD approaches have shown high effectiveness to successfully detect PAIs when materials used for fabrication these are known priori. However, most methods do not take into account characteristics PAIs' species in order generalise new, realistic and more challenging scenarios, where might be unknown....

Typically, real-world requirements to deploy face recognition models in unconstrained surveillance scenarios demand identify low-resolution faces with extremely low computational cost. In the last years, several methods based on complex deep learning have been proposed promising results but at a high Inspired by compactness and computation efficiency of lightweight networks their accuracy general tasks, this work we propose benchmark two recently introduced imagery enable efficient system...

One of the key factors to measure progress a research problem is design appropriate evaluation protocols defined on suitable databases. Recently, introduction comprehensive databases and benchmarks face videos has had great impact development new recognition techniques. However, most provided for these datasets are limited do not capture requirements unconstrained scenarios. That why sometimes performance methods current seems be saturated. To address this lack, tendency collect datasets,...

In spite of the advantages using fingerprints for subject authentication, several works have shown that fingerprint recognition systems can be easily circumvented by means artificial or presentation attack instruments (PAIs). order to address threat, existing detection (PAD) methods reported a high performance when materials used fabrication PAIs and capture devices are known. However, more complex realistic scenarios where one those factors remains unknown, these PAD unable correctly...

Artificial intelligence’s rapid advancement has enabled various applications, including intelligent video surveillance systems, assisted living, and human-computer interaction. These applications often require one core task: video-based human action recognition. Research in recognition is vast ongoing, making it difficult to assess the full scope of available methods current trends. This survey provides an in-depth exploration vision-based field, comprehensively offering techniques...

Training a model to recognize human actions in videos is computationally intensive. While modern strategies employ transfer learning methods make the process more efficient, they still face challenges regarding flexibility and efficiency. Existing solutions are limited functionality rely heavily on pretrained architectures, which can restrict their applicability diverse scenarios. Our work explores knowledge distillation (KD) for enhancing training of self-supervised video models three...

Parking block regions host dangerous behaviors that can be detected from a surveillance camera perspective. However, these are often occluded, subject to ground bumpiness or steep slopes, and thus they hard segment. Firstly, the paper proposes pyramidal solution takes advantage of satellite views same scene, based on deep Convolutional Neural Network (CNN). Training CNN perspective is rather impossible due combinatory explosion generated by multiple point-of-views. CNNs showed great promise...

Automatic evaluation of face image quality is an important topic in the development recognition systems. This paper describes a framework for assessing conformity images with parameter set standard ISO/IEC 19794-5 that determines if given possess identification value and can be used personal documents. New algorithms analysis classification respect to parameters evaluated proposed are presented. The proposal implemented as dynamic link library (DLL), offering many advantages use different...

Action recognition has been highlighted by its implications on issues of security and integrity. However, although the dense trajectory method achieved outstanding results, implementation is associated with a high computational cost, creating need for faster methods if we want to use these protect our safety integrity in real-time systems. In this work, explore subject keypoints estimation create what call action key trajectories, which version trajectories approach. We tested proposal KTH...

One of the main problems recognizing faces in videos is to achieve accurate algorithms which can be used real-time applications. Recently, Fisher Vector representation local descriptors (e.g., SIFT) has gained widespread popularity, achieving good recognition rates. In this work, we propose use encoding binary features for video face recognition, order speed up computation time representation. The experimental evaluation was conducted on challenging YouTube Faces database, showing that...