An enhanced real-time human pose estimation method based on modified YOLOv8 framework.

The objective of human pose estimation (HPE) derived from deep learning aims to accurately estimate and predict the human body posture in images or videos via the utilization of deep neural networks. However, the accuracy of real-time HPE tasks is still to be improved due to factors such as partial occlusion of body parts and limited receptive field of the model. To alleviate the accuracy loss caused by these issues, this paper proposes a real-time HPE model called CCAM - Person based on the YOLOv8 framework. Specifically, we have improved the backbone and neck of the YOLOv8x-pose real-time HPE model to alleviate the feature loss and receptive field constraints. Secondly, we introduce the context coordinate attention module (CCAM) to augment the model's focus on salient features, reduce background noise interference, alleviate key point regression failure caused by limb occlusion, and improve the accuracy of pose estimation. Our approach attains competitive results on multiple metrics of two open-source datasets, MS COCO 2017 and CrowdPose. Compared with the baseline model YOLOv8x-pose, CCAM-Person improves the average precision by 2.8% and 3.5% on the two datasets, respectively.

Characteristics of wild moose (Alces alces) vocalizations.

Moose are a popular species with recreationists but understudied acoustically. We used publicly available videos to characterize and quantify the vocalizations of moose in New Hampshire separated by age/sex class. We found significant differences in peak frequency, center frequency, bandwidth, and duration across the groups. Our results provide quantification of wild moose vocalizations across age/sex classes, which is a key step for passive acoustic detection of this species and highlights public videos as a potential resource for bioacoustics research of hard-to-capture and understudied species.

KARATECH: A Practice Support System Using an Accelerometer to Reduce the Preliminary Actions of Karate.

Kumite is a karate sparring competition in which two players face off and perform offensive and defensive techniques. Depending on the players, there may be preliminary actions (hereinafter referred to as "pre-actions"), such as pulling the arms or legs, lowering the shoulders, etc., just before a technique is performed. Since the presence of a pre-action allows the opponent to know the timing of the technique, it is important to reduce pre-actions in order to improve the kumite. However, it is difficult for beginners and intermediate players to accurately identify their pre-actions and to improve them through practice. Therefore, this study aims to construct a practice support system that enables beginners and intermediate players to understand their pre-actions. In this paper, we focus on the forefist punch, one of kumite's punching techniques. We propose a method to estimate the presence or absence of a pre-action based on the similarity between the acceleration data of an arbitrary forefist punch and a previously prepared dataset consisting of acceleration data of the forefist punch without a pre-action. We found that the proposed method can estimate the presence or absence of a pre-action in an arbitrary forefist punch with an accuracy of 86%. We also developed KARATECH as a system to support the practice of reducing pre-actions using the proposed method. KARATECH shows the presence or absence of pre-actions through videos and graphs. The evaluation results confirmed that the group using KARATECH had a lower pre-action rate.

Experiential-sensorial adaptive system model for hospitality based on 360° VR videos and case-based reasoning.

In this work, an adaptive software architecture is proposed for the generation of experiences for hotel promotion and marketing, based on Case-based Reasoning (CBR) that uses the attributes and user characteristics and immersive 360° videos. Considering that immersion in virtual reality (VR) environments can trigger responses in various dimensions, such as affective, cognitive, attitudinal, and behavioral dimensions, these dimensions are evaluated in immersive environments with 360° videos. To validate the results obtained with the software architecture, a quasi-experimental study was conducted through the evaluation of the experience, consisting in the visualization of the environments of a boutique hotel, with a sample of a randomly selected group of young people. The contribution of this work lies in the use of 360° VR videos, for the visualization of the hotel characteristics and environments according the user profiles, to evaluate the affective, cognitive and attitudinal and behavioral responses and their influence on the booking intention and attitude. Finally, conclusions and recommendations for future work have been established.

Remote Heart Rate Estimation Based on Transformer with Multi-Skip Connection Decoder: Method and Evaluation in the Wild.

Heart rate is an essential vital sign to evaluate human health. Remote heart monitoring using cheaply available devices has become a necessity in the twenty-first century to prevent any unfortunate situation caused by the hectic pace of life. In this paper, we propose a new method based on the transformer architecture with a multi-skip connection biLSTM decoder to estimate heart rate remotely from videos. Our method is based on the skin color variation caused by the change in blood volume in its surface. The presented heart rate estimation framework consists of three main steps: (1) the segmentation of the facial region of interest (ROI) based on the landmarks obtained by 3DDFA; (2) the extraction of the spatial and global features; and (3) the estimation of the heart rate value from the obtained features based on the proposed method. This paper investigates which feature extractor performs better by captioning the change in skin color related to the heart rate as well as the optimal number of frames needed to achieve better accuracy. Experiments were conducted using two publicly available datasets (LGI-PPGI and Vision for Vitals) and our own in-the-wild dataset (12 videos collected by four drivers). The experiments showed that our approach achieved better results than the previously published methods, making it the new state of the art on these datasets.

Highly dangerous road hazards are not immune from the low prevalence effect.

The low prevalence effect (LPE) is a cognitive limitation commonly found in visual search tasks, in which observers miss rare targets. Drivers looking for road hazards are also subject to the LPE. However, not all road hazards are equal; a paper bag floating down the road is much less dangerous than a rampaging moose. Here, we asked whether perceived hazardousness modulated the LPE. To examine this, we took a dataset in which 48 raters assessed the perceived dangerousness of hazards in recorded road videos (Song et al. in Behav Res Methods, 2023. https://doi.org/10.3758/s13428-023-02299-8 ) and correlated the ratings with data from a hazard detection task using the same stimuli with varying hazard prevalence rates (Kosovicheva et al. in Psychon Bull Rev 30(1):212-223, 2023. https://doi.org/10.3758/s13423-022-02159-0 ). We found that while hazard detectability increased monotonically with hazardousness ratings, the LPE was comparable across perceived hazardousness levels. Our findings are consistent with the decision criterion account of the LPE, in which target rarity induces a conservative shift in criterion. Importantly, feedback was necessary for a large and consistent LPE; when participants were not given feedback about their accuracy, the most dangerous hazards showed a non-significant LPE. However, eliminating feedback was not enough to induce the opposite of the LPE-prevalence induced concept change (Levari et al. in Science 360(6396):1465-1467, 2018. https://doi.org/10.1126/science.aap8731 ), in which participants adopt a more liberal criterion when instances of a category become rare. Our results suggest that the road hazard LPE may be somewhat affected by the inherent variability of driving situations, but is still observed for highly dangerous hazards.

Using visual attention estimation on videos for automated prediction of autism spectrum disorder and symptom severity in preschool children.

Atypical visual attention in individuals with autism spectrum disorders (ASD) has been utilised as a unique diagnosis criterion in previous research. This paper presents a novel approach to the automatic and quantitative screening of ASD as well as symptom severity prediction in preschool children. We develop a novel computational pipeline that extracts learned features from a dynamic visual stimulus to classify ASD children and predict the level of ASD-related symptoms. Experimental results demonstrate promising performance that is superior to using handcrafted features and machine learning algorithms, in terms of evaluation metrics used in diagnostic tests. Using a leave-one-out cross-validation approach, we obtained an accuracy of 94.59%, a sensitivity of 100%, a specificity of 76.47% and an area under the receiver operating characteristic curve (AUC) of 96% for ASD classification. In addition, we obtained an accuracy of 94.74%, a sensitivity of 87.50%, a specificity of 100% and an AUC of 99% for ASD symptom severity prediction.

Correlation between pose estimation features regarding movements towards the midline in early infancy.

In infants, spontaneous movement towards the midline (MTM) indicates the initiation of anti-gravity ability development. Markerless 2D pose estimation is a cost-effective, time-efficient, and quantifiable alternative to movement assessment. We aimed to establish correlations between pose estimation features and MTM in early-age infants. Ninety-four infant videos were analysed to calculate the percentage and rate of MTM occurrence. 2D Pose estimation processed the videos and determined the distances and areas using wrist and ankle landmark coordinates. We collected data using video recordings from 20 infants aged 8-16 weeks post-term age. Correlations between MTM observations and distance values were evaluated. Differences in areas between groups of videos showing MTM and no MTM in the total, lower-limb, and upper-limb categories were examined. MTM observations revealed common occurrences of hand-to-trunk and foot-to-foot movements. Weak correlations were noted between limb distances to the midbody imaginary line and MTM occurrence values. Lower MTM showed significant differences in the lower part (p = 0.003) and whole area (p = 0.001). Video recording by parents or guardians could extract features using 2D pose estimation, assisting in the early identification of MTM in infants. Further research is required to assess a larger sample size with the diversity of MTM motor behaviour, and later developmental skills, and collect data from at-risk infants.

Individual Activity Anomaly Estimation in Operating Rooms Based on Time-Sequential Prediction.

Surveillance videos of operating rooms have potential to benefit post-operative analysis and study. However, there is currently no effective method to extract useful information from the long and massive videos. As a step towards tackling this issue, we propose a novel method to recognize and evaluate individual activities using an anomaly estimation model based on time-sequential prediction. We verified the effectiveness of our method by comparing two time-sequential features: individual bounding boxes and body key points. Experiment results using actual surgery videos show that the bounding boxes are suitable for predicting and detecting regional movements, while the anomaly scores using key points can hardly be used to detect activities. As future work, we will be proceeding with extending our activity prediction for detecting unexpected and urgent events.

Efficient Human Violence Recognition for Surveillance in Real Time.

Human violence recognition is an area of great interest in the scientific community due to its broad spectrum of applications, especially in video surveillance systems, because detecting violence in real time can prevent criminal acts and save lives. The majority of existing proposals and studies focus on result precision, neglecting efficiency and practical implementations. Thus, in this work, we propose a model that is effective and efficient in recognizing human violence in real time. The proposed model consists of three modules: the Spatial Motion Extractor (SME) module, which extracts regions of interest from a frame; the Short Temporal Extractor (STE) module, which extracts temporal characteristics of rapid movements; and the Global Temporal Extractor (GTE) module, which is responsible for identifying long-lasting temporal features and fine-tuning the model. The proposal was evaluated for its efficiency, effectiveness, and ability to operate in real time. The results obtained on the Hockey, Movies, and RWF-2000 datasets demonstrated that this approach is highly efficient compared to various alternatives. In addition, the VioPeru dataset was created, which contains violent and non-violent videos captured by real video surveillance cameras in Peru, to validate the real-time applicability of the model. When tested on this dataset, the effectiveness of our model was superior to the best existing models.

Multiple mask and boundary scoring R-CNN with cGAN data augmentation for bladder tumor segmentation in WLC videos.

Automatic diagnosis systems capable of handling multiple pathologies are essential in clinical practice. This study focuses on enhancing precise lesion localization, classification and delineation in transurethral resection of bladder tumor (TURBT) to reduce cancer recurrence. Despite deep learning models success, medical applications face challenges like small and limited datasets and poor image characterization, including the absence lack of color/texture modeling. To address these issues, three solutions are proposed: (1) an improved texture-constrained version of the pix2pixHD cGAN for data augmentation, addressing the tradeoff of generating high-quality images with enough stochasticity using the Fréchet Inception Distance (FID) measure. (2) Introducing the Multiple Mask and Boundary Scoring R-CNN (MM&BS R-CNN), a new mask sub-net scheme where multiple masks are generated from the different levels of the mask sub-net pipeline, improving segmentation accuracy by including a new scoring module to refine object boundaries. (3) A novel accelerated training strategy based on the SGD optimizer with the second momentum. Experimental results show significant mAP improvements: the data generation scheme improves by more than 12 %; MM&BS R-CNN proposed architecture is responsible for an improvement of about 1.25 %, and the training algorithm based on the second-order momentum increases mAP by 2-3 %. The simultaneous use of all three proposals improved the state-of-the-art mAP by 17.44 %.

An innovative method for human height estimation combining video images and 3D laser scanning.

Digitalization has increased the number of video surveillance systems that sometimes capture crime images. Traditional methods of human height estimation use projective geometry. However, sometimes they cannot be used because the video camera surveillance system is not available or has been moved and there are no reference lines on the frame. Scientific studies have developed a new method for human height estimation using 3D laser scanning. This model necessarily requires a series of approximations, which increase the final measurement error. To overcome this problem, in the present study, images of a subject are projected directly on the 3D model, estimating the height of the subject. This article describes the methodological approach adopted through the analysis of a real case study in a controlled environment executed by Carabinieri Forensic Investigation Department (Italy). The aim is to obtain a human anthropometric measure derived from frames extracted from the videos associated with the digital survey of the framed area obtained with 3D laser scanning and point cloud analysis. The result is the height estimation of five subjects filmed by a camera obtained through the combination of 2D images extracted by a DVR/surveillance systems with 3D laser scanning. Results show that most estimated measurements are less than the real measurement of the subject; it also depends on the posture of the subject while walking. Furthermore, results shows the differences between the real height and the estimated height with a statistical approach.

Guidelines for the Use of Generative Artificial Intelligence Tools for Biomedical Journal Authors and Reviewers.

Authors are permitted to use generative artificial intelligence (AI) large language models (LLM) to improve the readability of their own writing. However, authors must review and edit the output resulting from generative AI and are accountable for the accuracy of their publications. AI may not be listed, or cited, as an author. Authors who use AI in the scientific writing process must disclose the use of AI LLM in their manuscript including a description of the tool and reason for use. Authors are not permitted to use AI to create or alter images or videos, (unless this is part of the research design in which case a statement is required explaining what was created or altered, with what tools, how, and for what reason). Finally, AI use by reviewers and editors is not permitted and violates confidentiality and proprietary rights and may breach data privacy rights. In conclusion, scientific writing and peer review is the responsibility of humans.

Preschool teachers' knowledge, beliefs, and self-efficacy concerning autism: A parallel mixed-methods study of an intervention to improve autism identification.

LAY ABSTRACT: Preschool teachers can play a critical role in early detection of autism. Equipping preschool teachers with prerequisite knowledge and skills would allow them to identify children with probable autism and referral to diagnostic services. This study aimed to investigate the impact of an educational module (EMiASD) that prepared preschool teachers to identify autism symptoms. The sample included 144 preschool teachers, of which 120 were stratified and randomly assigned to an intervention arm receiving training in EMiASD (n = 60) or a comparison arm receiving standard training (n = 60) using a parallel mixed-methods design. Responses to open-ended questions about video case studies revealed improvement in the identification of autism symptoms in preschool teachers in the intervention arm, in contrast to preschool teachers in the comparison arm. Moreover, significant changes in knowledge, belief, and self-efficacy about autism favoured EMiASD. Overall, these results demonstrate the influence of EMiASD in the Yemeni cultural context.

Development of convolutional neural network models that recognize normal anatomic structures during real-time radial-array and linear-array EUS (with videos).

BACKGROUND AND AIMS: EUS is a high-skill technique that requires numerous procedures to achieve competence. However, training facilities are limited worldwide. Convolutional neural network (CNN) models have been previously implemented for object detection. We developed 2 EUS-based CNN models for normal anatomic structure recognition during real-time linear- and radial-array EUS evaluations. METHODS: The study was performed from February 2020 to June 2022. Consecutive patient videos of linear- and radial-array EUS videos were recorded. Expert endosonographers identified and labeled 20 normal anatomic structures within the videos for training and validation of the CNN models. Initial CNN models (CNNv1) were developed from 45 videos and the improved models (CNNv2) from an additional 102 videos. CNN model performance was compared with that of 2 expert endosonographers. RESULTS: CNNv1 used 45,034 linear-array EUS frames and 21,063 radial-array EUS frames. CNNv2 used 148,980 linear-array EUS frames and 128,871 radial-array EUS frames. Linear-array CNNv1 and radial-array CNNv1 achieved a 75.65% and 71.36% mean average precision (mAP) with a total loss of .19 and .18, respectively. Linear-array CNNv2 obtained an 88.7% mAP with a .06 total loss, whereas radial-array CNNv2 achieved an 83.5% mAP with a .07 total loss. CNNv2 accurately detected all studied normal anatomic structures with a >98% observed agreement during clinical validation. CONCLUSIONS: The proposed CNN models accurately recognize the normal anatomic structures in prerecorded videos and real-time EUS. Prospective trials are needed to evaluate the impact of these models on the learning curves of EUS trainees.

How Do Physicians Frame Medical Information in Talks With Their Patients? An Inductive Microanalysis.

During medical consultations, physicians need to share a substantial amount of information with their patients. How this information is framed can be crucial for patient understanding and outcomes, but little is known about the details of how physicians frame information in practice. Using an inductive microanalysis approach in the study of videotaped medical interactions, we aimed to identify the information frames (i.e., higher-level ways of organizing and structuring information to reach a particular purpose) and the information-framing devices (i.e., any dialogic mechanism used to present information in a particular way that shapes how the patient might perceive and interpret it) physicians use spontaneously and intuitively while sharing information with their patients. We identified 66 different information-framing devices acting within nine information frames conveying: (1) Do we agree that we share this knowledge?, (2) I don't like where I (or where you are) am going with this, (3) This may be tricky to understand, (4) You may need to think, (5) This is important, (6) This is not important, (7) This comes from me as a doctor, (8) This comes from me as a person, and (9) This is directed to you as a unique person. The kaleidoscope of information-framing devices described in this study reveals the near impossibility for neutrality and objectivity in the information-sharing practice of medical care. It also represents an inductively derived starting point for further research into aspects of physicians' information-sharing praxis.

Classifying Vocal Folds Fixation from Endoscopic Videos with Machine Learning.

Vocal folds motility evaluation is paramount in both the assessment of functional deficits and in the accurate staging of neoplastic disease of the glottis. Diagnostic endoscopy, and in particular videoendoscopy, is nowadays the method through which the motility is estimated. The clinical diagnosis, however, relies on the examination of the videoendoscopic frames, which is a subjective and professional-dependent task. Hence, a more rigorous, objective, reliable, and repeatable method is needed. To support clinicians, this paper proposes a machine learning (ML) approach for vocal cords motility classification. From the endoscopic videos of 186 patients with both vocal cords preserved motility and fixation, a dataset of 558 images relative to the two classes was extracted. Successively, a number of features was retrieved from the images and used to train and test four well-grounded ML classifiers. From test results, the best performance was achieved using XGBoost, with precision = 0.82, recall = 0.82, F1 score = 0.82, and accuracy = 0.82. After comparing the most relevant ML models, we believe that this approach could provide precise and reliable support to clinical evaluation.Clinical Relevance- This research represents an important advancement in the state-of-the-art of computer-assisted otolaryngology, to develop an effective tool for motility assessment in the clinical practice.

Segmenting Cardiac Ultrasound Videos Using Self-Supervised Learning.

Deep learning models trained with an insufficient volume of data can often fail to generalize between different equipment, clinics, and clinicians or fail to achieve acceptable performance. We improve cardiac ultrasound segmentation models using unlabeled data to learn recurrent anatomical representations via self-supervision. In addition, we leverage supervised local contrastive learning on sparse labels to improve the segmentation and reduce the need for large amounts of dense pixel-level supervisory annotations. Then, we implement supervised fine-tuning to segment key temporal anatomical features to estimate the cardiac Ejection Fraction (EF). We show that pretraining the network weights using self-supervised learning for subsequent supervised contrastive learning outperforms learning from scratch, validated using two state-of-the-art segmentation models, the DeepLabv3+ and Attention U-Net.Clinical relevance-This work has clinical relevance for assisting physicians when conducting cardiac function evaluations. We improve cardiac ejection fraction evaluation compared to previous methods, helping to alleviate the burden associated with acquiring labeled images.

A Full Transformer-based Framework for Automatic Pain Estimation using Videos.

The automatic estimation of pain is essential in designing an optimal pain management system offering reliable assessment and reducing the suffering of patients. In this study, we present a novel full transformer-based framework consisting of a Transformer in Transformer (TNT) model and a Transformer leveraging cross-attention and self-attention blocks. Elaborating on videos from the BioVid database, we demonstrate state-of-the-art performances, showing the efficacy, efficiency, and generalization capability across all the primary pain estimation tasks.

A single session of action observation therapy versus observing a natural landscape in adults with chronic neck pain - a randomized controlled trial.

BACKGROUND: Action observation (AO) has emerged as a potential neurorehabilitation therapy for patients with neck pain (NP), but evidence of its effectiveness is scarce. This study aims to assess the effect of a single session of AO when compared to observing a natural landscape on NP intensity, fear of movement, fear-avoidance beliefs, neck muscles' strength, pressure pain threshold, and tactile acuity. METHODS: Sixty participants with NP were randomly allocated to the AO group (n = 30) or control group (n = 30). Both groups watched an 11-minute video: the AO group watched a video of a person matched for age and sex performing neck exercises, while the control group watched a video of natural landscapes. Neck pain intensity, fear of movement, fear-avoidance beliefs, tactile acuity, pressure pain thresholds, and neck muscle strength were assessed both at baseline and post-intervention. General linear models of repeated measures (ANCOVA of two factors) were used to explore between-group differences at post-intervention. RESULTS: There was a significant main effect of time for pain intensity (p = 0.02; η2p = 0.09; within-group mean change and 95% CI: AO=-1.44 (-2.28, -0.59); control=-1.90 (-2.74, -1.06), but no time versus group interaction (p = 0.46). A time versus group significant interaction was found for one out of the six measurement sites of two-point discrimination and the neck flexors strength (p < 0.05) favoring the control group. No other statistically significant differences were found for the remaining variables). CONCLUSIONS: Results suggest a similar acute benefit for both a single session of AO and observing natural landscapes for promoting hypoalgesia, but no impact on kinesiophobia, fear-avoidance beliefs, or pressure pain thresholds. Also, AO had no positive effect on two-point discrimination and muscle strength. Further research is needed, with longer interventions. TRIAL REGISTRATION: Clinialtrials.gov (NCT05078489).