Search Results (336)

Two-dimensional (2D) human pose estimation is one of the key research directions in Computer Vision (CV), which has wide application prospects in behavior recognition, such as gesture tracking, intelligent monitoring, and identity recognition. Therefore, it has recently attracted extensive attention from academia and industry. However, although a large amount of literature has been published, existing reviews often lack a unified theoretical perspective and fail to capture the latest paradigm shifts brought by foundation models. To this end, this paper reviews the applications of deep learning in the domain of 2D body pose estimation from 2010 to 2025 through a cascading approach. First, the mainstream body pose datasets and related evaluation metrics are introduced in a comprehensive and convincing way through mathematical formulas. Subsequently, an in-depth analysis of the performance of the algorithms in single-person and multi-person scenarios, and a comprehensive comparative analysis of the strengths and weaknesses of each algorithmic model, are conducted. A comprehensive comparative analysis encompassing both traditional architectures and the latest deep learning breakthroughs are provided, specifically highlighting Vision Foundation Models (VFMs), generative Diffusion processes, and State Space Models (SSMs). Finally, the current state of research in the field of 2D human pose estimation is summarized, and three main challenges, emerging solutions, and expected development trends are pointed out. This survey is an exhaustive compilation of existing research in 2D human pose estimation, providing a blueprint for researchers in the field and laying the foundation for future research work. Full article

Traditional acupoint localization methods rely heavily on manual operation, resulting in high subjectivity and limited accuracy. To improve the precision and stability of acupoint detection, this study integrates machine vision technology with in situ projection to achieve automated recognition and real-time visualization of human acupoints. First, an automatic calibration method based on image processing is proposed for back acupoints. Spinal features are extracted from the blue channel, enhanced using adaptive histogram equalization, and processed through region of interest extraction, minimum-threshold binarization, and morphological operations. Key spinal curve points are then fitted using Bézier functions. Canny edge detection is used to extract the human silhouette, locate the acromion, and derive the pixel scale of the “cun” measurement, enabling coordinate computation for 141 back acupoints. In the deep learning component, an improved YOLOv8-Pose model is developed for acupoint localization. Unlike existing methods that use local attention or the original Object Keypoint Similarity (OKS) loss, we introduce two innovations: a non-local attention module for global dependency modeling, and a novel Efficient Object Keypoint Similarity (EOKS) loss function that incorporates geometric constraints—namely, width, height, and center distance—in addition to Euclidean distance. A non-local attention mechanism is incorporated into the backbone to enhance global feature extraction, and the EOKS loss function is designed to improve spatiogeometric regression accuracy. An inference mechanism is further introduced to derive the remaining acupoints from 49 detected keypoints; experiments demonstrate that the improved model achieves 95.0% detection accuracy, outperforming the baseline by 2.62%, with an inference time of 14.5 ms. Finally, an in situ projection platform is constructed, combining camera calibration, four-point proportional scaling, and an OpenCV 4.5.4-based interactive interface. The system supports real-time translation, rotation, and scaling, enabling accurate projection of detected acupoints onto the human body. Full article

Vision Transformer (ViT) has been widely adopted in the computer vision community. However, the standard ViT often contains many parameters, usually performs poorly when trained from scratch on medium-scale datasets, and does not explicitly preserve the local spatial and channel-wise structures within each token. This work proposes a novel model called the Token-Shared Convolutional Projection Vision Transformer (TSCP-ViT). The core idea of TSCP-ViT is to integrate convolutional layers into the multi-head attention mechanism and to apply the same convolutional operation independently to each token, where each token exhibits spatial 2D multi-channel characteristics. In addition, this work introduces a Transformer decoder immediately after each Transformer encoder, enabling the classification tokens to aggregate information from all tokens and be updated using statistical information. Moreover, a trainable Non-Reversing Gate GELU (NRG-GELU) activation is also proposed. Comparative experiments on CIFAR-100, Food-101, and ImageNet100 show that, under comparable parameter counts and without pretraining or knowledge distillation, TSCP-ViT substantially surpasses ViT, outperforms CvT, outperforms ResNet on Food-101, and approaches ResNet on CIFAR-100 and ImageNet100, although with considerably higher FLOPs. Full article

Computer vision (CV) has significantly advanced in object detection and multi-object tracking; however, its application to modeling safety-critical social behaviors for blind and low-vision (BLV) individuals remains limited. In particular, sustained behaviors such as stalking—characterized by persistent proximity and trajectory consistency—have not been systematically addressed within wearable assistive systems. To investigate this gap, we first conducted a formative user study combining semi-structured interviews and behavioral observations to identify safety concerns and wearable design requirements among BLV participants. The findings reveal recurring concerns regarding prolonged following behaviors and highlight the importance of privacy-preserving, socially unobtrusive device configurations. Guided by these insights, we develop a shoulder-slung wearable system integrating dual-camera sensing with an edge-based vision processing pipeline. We reformulate stalking detection as a temporal behavioral persistence problem built upon multi-object tracking (MOT). Leveraging FairMOT for identity-preserving tracking and monocular depth estimation for spatial modeling, we introduce an online temporal persistence-based risk scoring mechanism that accumulates proximity and directional consistency over time. The complete pipeline operates in real time on an embedded platform without cloud dependency. By bridging user-centered design and behavior-oriented visual inference, this work demonstrates how MOT outputs can be extended beyond identity preservation to support temporally coherent safety assessment in wearable assistive contexts. Full article

This study proposes a real-time multi-factor behavioral monitoring framework for Computer Vision Syndrome (CVS) using computer vision techniques and TensorFlow for browser-based implementation. Four vision-based detection pipelines—Dlib-based, MediaPipe-based, CNN-based, and TensorFlow-based implementations—were evaluated to identify a suitable configuration for real-time deployment. The selected browser-based implementation integrated MediaPipeFaceMesh for facial landmark extraction and MoveNet SinglePose Lightning for supplementary pose-related detection. During the pipeline-selection stage, the Dlib-based pipeline showed high task-specific accuracy in blink detection (0.9034) and head pose estimation (0.9005), while the MediaPipe-based pipeline provided the highest processing speed for these tasks, with 73.09 FPS and 75.36 FPS, respectively. The CNN-based baseline showed limited real-time suitability, with low F1-scores and FPS values ranging from 4.22 to 7.32 across tasks. These preliminary comparison results informed the selection of the browser-based pipeline, which provided the most practical trade-off among detection performance, real-time processing capability, browser-based execution, and deployment flexibility. In blink detection, the selected pipeline achieved a precision of 0.8906, a recall of 0.9490, an F1-score of 0.9189, and 13.94 FPS. The proposed framework integrates five core operational indicators: viewing distance, vertical viewing deviation, horizontal viewing deviation, blink rate, and continuous usage duration. These indicators support rule-based real-time alerts and session-based behavioral pattern analysis. After implementation, the prototype operated in real time, detected concurrent CVS-related behavioral conditions, generated interpretable rule-based alerts, and summarized recurring behavioral patterns across a monitoring session. A controlled alert-level evaluation further indicated that the warning layer operated consistently for most rule-based alert conditions, although low-blink and prolonged-focus alerts require further refinement. These findings highlight the potential of combining browser-based visual detection with interpretable operational indicators for practical CVS-related behavioral monitoring. Full article

Regulations in Poland require acceptance load tests to verify bridge response under moving loads before structures are approved for operation. These tests are mandatory for new bridges, after major renovations, and for reconstructed structures, and may also be conducted as supplementary assessments of existing bridges to determine their load-carrying capacity. This paper presents one of the first documented applications, to the authors’ knowledge, of low-cost sensing technology for capturing bridge displacements with sub-millimeter tracking precision during acceptance load testing. The study explores the use of modern remote sensing methods based on digital image correlation (DIC) to assess vertical displacements of a truss railway bridge span under moving loads. Video data were recorded using a standard smartphone under nighttime conditions with artificial lighting, demonstrating a highly accessible and cost-effective measurement approach. The collected data were processed using the DES Vision System and compared with results obtained from traditional measurement techniques, such as accelerometers, enabling an evaluation of the accuracy and precision of the DIC method. The findings show that smartphone-based video recordings can provide displacement measurements with millimeter- to sub-millimeter-level tracking precision. Additionally, a numerical finite element method (FEM) model was developed to support interpretation of the structural response under moving loads. Full article

The accelerating deterioration of Chinese historical villages necessitates advanced digital approaches for systematic documentation and conservation. The present research proposes a novel Digital Heritage Framework that integrates UAV-based 3D oblique photogrammetry, LiDAR point cloud modeling, and computer vision. Unlike single-technology approaches, our methodology solves modeling issues for complex terrain mapping. This especially applies to the interior and roof works of buildings. The framework implements a customized Rhino-Grasshopper. The 3D model is able to resolve issues of shadow occlusion and spatial discontinuity by integrating aerial and ground-based datasets into spatially coherent formats. This makes use of the Meta-AI-SAM2 deep learning model for semantic segmentation and identification of materials. The computer vision (CV) approach gives semi-automated façade analysis. It enables documentation of complex architectural features non-invasively. We developed a Unity-based visualization platform. It features multiscale representations, ranging from village-scale layouts to centimeter-accurate scans of heritage structures such as the Qinchuan Ancestral Hall. Integration with the Unity platform optimizes dataset organization and hierarchical structuring. This significantly enhances database operational efficiency. This integration reduces manual processing complexity and hardware demands. Demonstrating documented efficiency and precision, this workflow presents a scalable solution for endangered heritage sites. Future research will explore AI-assisted detail reconstruction and cross-cultural adaptations. It potentially establishes this framework as a comprehensive tool for sustainable digital conservation. Full article

Crackers are a popular and convenient snack; however, the baking process can produce neo-formed contaminants (NFCs), such as acrylamide (AA) and 5-hydroxymethylfurfural (HMF), through non-enzymatic browning reactions. Conventional analytical methods for quantifying these NFCs are complex, labor-intensive, and require specialized personnel. The main objective of this study was to develop computer vision (CV) models based on surface digital image analysis for the rapid prediction of AA and HMF in crackers. Therefore, five baking temperatures (160, 170, 180, 190, and 200 °C) and times (15, 20, 25, 30, and 35 min) were tested and analyzed using CV alongside conventional analytical methods. CV estimates and analytical measurements for AA (4.35–829 µg kg⁻¹) and HMF (0.004–105.4 mg kg⁻¹) contents were compared using cross-validation with a “leave-one-treatment-out” approach. The average error for missing measurements was 3.10% for AA and 3.28% for HMF, validating CV as an efficient tool for the rapid estimation of these NFCs in crackers. Among the cracker samples with AA content below the EU benchmark of 400 μg/kg, evaluated using the Check-All-That-Apply (CATA) test, consumers preferred the samples baked at 180 °C for 25 min, which also exhibited the lowest levels of both AA and HMF. Full article

Artificial Intelligence (AI) is revolutionizing Construction Management (CM) through automation, predictive analytics, and real-time decision-making throughout the project lifecycle.This study aims to provide a comprehensive and structured synthesis of AI models and their applications in CM. This paper presents a systematic review of 191 peer-reviewed articles published between 2020 and 2025, aiming to integrate the current state of AI implementation in CM, focusing on AI methods and models and their applications in CM. Compared to previous reviews that take these factors individually or focus narrowly on specific techniques, this study offers a comprehensive taxonomy that systematically maps AI techniques against CM functions and integration platforms. The results reveal that AI applications are primarily concentrated in risk and safety management, decision support, and monitoring and control, while domains such as legal analytics, robotics, and cybersecurity remain underexplored. Furthermore, Computer Vision (CV) and Deep Learning (DL) dominate tasks such as safety monitoring and defect detection, whereas Machine Learning (ML) and optimization algorithms are widely applied in cost estimation and scheduling. It also addresses developments rarely covered in construction research, including Generative AI (Gen-AI), Explainable AI (XAI), and transformer models, presenting a strategic framework for the widespread adoption of AI in the construction environment. This study contributes a structured taxonomy that systematically links AI models with CM functions and enabling technologies, providing a comprehensive synthesis of emerging trends and research gaps. Full article

Machine learning techniques are increasingly used for computer vision applications in manufacturing. Synthetic data, generated through realistic simulations, are utilized to accelerate the data collection process while optimizing accuracy and precision of ML models. However, in manufacturing there is usually the need for the development of several CV applications that support different production steps. This obstacle requires a systematic approach for generating synthetic datasets that can be used for developing effective CV systems. Hence, this work presents a pipeline for generating photorealistic synthetic datasets, using a set of digital tools such as 3D modeling, photorealistic rendering, automated labeling, and ML training tools. The proposed framework is tested and validated in a robot-assisted packaging case in the dairy industry. The industrial use case provides a pilot-level demonstration that the synthetic dataset generation framework can support the development of CV modules across several production steps and thus it can aid in accelerating commissioning and reconfiguration of industrial automation setups. Moreover, the pilot validation indicates that object detection and recognition models trained on synthetic data can provide sufficient performance for the specific requirements of the examined packaging scenario. Full article

Identification of unknown individuals is challenging, and radiological imaging databases provide rich anatomical information for automated recognition. This study evaluated whether a single routine magnetic resonance imaging (MRI) slice contains sufficient person-specific features to identify individuals in large databases. It analyzed 11,078 head MRI examinations from 5770 individuals (age 52 ± 18 years, 2714 men) acquired between 2002 and 2025. For identification, 112 individuals were randomly selected across eight 10-year age groups, and one slice from four anatomical regions was extracted. The remaining 10,966 MRI examinations with 247,804 slices formed the reference database. Distinctive anatomical features were automatically extracted using computer vision (CV), and the identification rate was evaluated by rank. Using a single MRI slice, the identification rate at rank 1 reached 96% (107/112) for the best-performing region, the maxillary sinus, among 5770 potential identities. Across all regions, the rank 1 identification rate ranged from 91% to 96%; combining them increased rank 1 and 10 identification rates to 98% (110/112) and 99% (111/112). Identification rate remained stable over several years, with only two cases showing reduced rank 1 performance, likely due to age-related morphological changes. A single MRI slice contains stable, individualized features sufficient for reliable identification in large databases, supporting automated CV-based personal identification across years. Full article

Open-pit mining operations rely heavily on visual inspection to identify indicators of slope instability such as surface cracks. Early identification of these geotechnical hazards enables timely safety interventions to protect both workers and assets in the event of slope failures or landslides. While computer vision (CV) approaches offer a promising avenue for autonomous crack detection, their effectiveness remains constrained by the scarcity of labelled geotechnical datasets. Deep learning (DL)-based models, in particular, require large amounts of representative training data to generalize to unseen conditions; however, collecting such data from operational mine sites is limited by safety, cost, and data confidentiality constraints. To address this challenge, this study proposes a novel hybrid game engine–generative artificial intelligence (AI) framework for large-scale dataset generation without requiring real-world training data. Leveraging a parameterized virtual environment developed in Unreal Engine 5 (UE5), the framework generates realistic images of open-pit surface cracks and enhances their fidelity and diversity using StyleGAN2-ADA. The synthesized datasets were used to train the YOLOv11 real-time object detection model and evaluated on a held-out real-world dataset of open-pit slope imagery to assess the effectiveness of the proposed framework in improving model generalizability under extreme data scarcity. Experimental results demonstrated that models trained using the proposed framework consistently outperformed the UE5 baseline, with average precision (AP) at intersection over union (IoU) thresholds of 0.5 and [0.5:0.95] increasing from 0.792 to 0.922 (+16.4%) and 0.536 to 0.722 (+34.7%), respectively, across the best-performing configurations. These findings demonstrate the effectiveness of hybrid generative AI frameworks in mitigating data scarcity in CV applications and supporting the development of scalable automated slope monitoring systems for improved worker safety and operational efficiency in open-pit mining. Full article

Artificial intelligence (AI) is rapidly reshaping cardiovascular (CV) medicine, driving a paradigm shift toward truly personalized and data-driven care. This comprehensive review examines the conceptual foundations, clinical applications, and future implications of AI across the CV continuum, spanning prevention, diagnosis, risk stratification, and therapy. Core AI methodologies (including machine learning, deep learning, natural language processing, and computer vision) are discussed in the context of cardiology’s uniquely data-rich environment, encompassing imaging, electrocardiography, electronic health records, wearable devices, and multi-omics data. This systematic review highlights major clinical domains where AI has demonstrated a substantial impact, including CV imaging, ECG interpretation, hypertension and heart failure management, coronary artery disease, acute coronary syndromes, interventional cardiology, and cardiac surgery. AI-driven predictive analytics enable early detection of subclinical disease, improved prognostication, and individualized prevention strategies, while wearable technologies and remote monitoring platforms facilitate continuous, real-world patient surveillance. Emerging applications in pharmacotherapy, drug repurposing, and genomics further reinforce AI’s role in advancing precision cardiology. Equally emphasized are the ethical, legal, and social challenges accompanying AI adoption, such as algorithmic bias, data privacy, cybersecurity, interpretability, and regulatory oversight. Our review underscores the necessity of rigorous clinical validation, transparent model design, and seamless integration into clinical workflows to ensure safety, equity, and physician trust. Ultimately, AI is best positioned as an augmentative tool that complements (but does not replace!) clinical expertise. By fostering hybrid intelligence that integrates human judgment with computational power, AI has the potential to redefine CV care delivery, improve outcomes, and support a more proactive, patient-centered healthcare model. Full article

Agonistic behaviors such as aggression, ear biting, and tail biting remain major challenges for pig welfare, particularly during the weaning and growing periods. Computer vision (CV) technologies are emerging as scalable tools for non-invasive monitoring of these behaviors. This systematic review summarizes recent advances in CV-based detection of agonistic behaviors in pigs and identifies factors influencing their reliability and commercial adoption. Following the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines, a structured search of Scopus, Web of Science, and PubMed identified 42 eligible studies. Most studies employ deep learning approaches, including you only look once (YOLO)-based detectors and spatio-temporal models, achieving detection accuracy of up to 97% for behaviors such as head knocking, head-to-body pushing, and tail biting, typically evaluated under controlled conditions using mAP@0.5. Three key findings emerged: rapid progress in deep learning-based detection; methodological heterogeneity in behavioral definitions, validation strategies, and annotation protocols; and a gap between high detection accuracy and demonstrated improvements in welfare or productivity. Progress is limited by scarce cross-farm validation, inconsistent bout definitions, reliance on manual annotations, and weak integration with physiological and production indicators. Future research should prioritize standardized behavioral definitions, multimodal integration, predictive modeling, and rigorous external validation. Full article

Social media platforms have become primary communication channels for technical European universities. However, the extent to which global platform algorithms homogenize individual preferences across cultures remains underexplored. Although the current literature offers insights into the topic, none of the works consider the cross-national and multimodal nature of the phenomenon. In the current paper, we introduce the Media Information Processing System (MIPS), a privacy-preserving multimodal deep learning (DL) framework that incorporates large language models (LLMs), computer vision (CV), and knowledge graphs. We analyze data from 15,520 public videos shared by 2359 followers of six top technical universities from Romania, Germany, Italy, and Russia. The results of the study suggest that the degree of homogeneity of the followers’ interest profiles is markedly high. Statistical profiling of the data indicates that the interest profiles of the followers from different countries are positively correlated with a high degree of strength (mean Pearson r = 0.96; p > 0.90). Consensus clustering of the data reveals the existence of stable clusters of themes with high stability scores (>0.75), such as “Human Interaction Dynamics”. The results of the study contradict the traditional theory of regional cultural differentiation. Instead, the results suggest the existence of a new “digital student persona” that is characteristic of the academic lifestyle of students from different countries. Full article