Search

Affective computing has suffered by the precise annotation because the emotions are highly subjective and vague. The music video emotion is complex due to the diverse textual, acoustic, and visual information which can take the form of lyrics, singer voice, sounds from the different instruments, and visual representations. This can be one reason why there has been a limited study in this domain and no standard dataset has been produced before now. In this study, we proposed an unsupervised method for music video emotion analysis using music video contents on the Internet. We also produced a labelled dataset and compared the supervised and unsupervised methods for emotion classification. The music and video information are processed through a multimodal architecture with audio-video information exchange and boosting method. The general 2D and 3D convolution networks compared with the slow-fast network with filter and channel separable convolution in multimodal architecture. Several supervised and unsupervised networks were trained in an end-to-end manner and results were evaluated using various evaluation metrics. The proposed method used a large dataset for unsupervised emotion classification and interpreted the results quantitatively and qualitatively in the music video that had never been applied in the past. The result shows a large increment in classification score using unsupervised features and information sharing techniques on audio and video network. Our best classifier attained 77% accuracy, an f1-score of 0.77, and an area under the curve score of 0.94 with minimum computational cost., (© 2021. The Author(s).)

Objective: The operational definition of interictal epileptiform discharges (IEDs) of the International Federation of Clinical Neurophysiology (IFCN) described six morphological criteria. Our objective was to assess the impact of pattern-repetition in the EEG-recording, on the diagnostic accuracy of using the IFCN criteria. For clinical implementation, specificity over 95% was set as target., Methods: Interictal EEG-recordings of 20-minutes, containing sharp-transients, from 60 patients (30 with epilepsy and 30 with non-epileptic paroxysmal events) were evaluated by three experts, who first marked IEDs solely based on expert opinion, and then, independently from the first session evaluated the presence of the IFCN criteria for each sharp-transient. The gold standard was derived from long-term video-EEG recordings of the patientś habitual paroxysmal episodes., Results: Presence of at least one discharge fulfilling five criteria provided a specificity of 100% (sensitivity: 70%). For discharges fulfilling fewer criteria, a higher number of discharges was needed to keep the specificity over 95% (5 discharges, when only 3 criteria were fulfilled). A sequential combination of these sets of criteria and thresholds provided a specificity of 97% and sensitivity of 80%., Conclusions: Pattern-repetition and IED morphology influence diagnostic accuracy., Significance: Systematic application of these criteria will improve quality of clinical EEG interpretation., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2021 International Federation of Clinical Neurophysiology. Published by Elsevier B.V. All rights reserved.)

Objectives: The purpose of our study is to evaluate the information available on YouTube regarding FGCS procedures in terms of content, accuracy, reliability and quality., Methods: Videos were listed on YouTube on 01.10.2020 by searching the terms "plastic surgery for female genital", "vaginoplasty", "labiaplasty" and selecting their relevance from the filtering options. Among these listed videos, the 50 most watched videos were listed and recorded excluding commercials, repeating videos, those shorter than 60s and non-English ones. The videos were scored by two researchers using DISCERN and JAMA scales. The DISCERN scale varies between 0-75 points and according to the scores obtained from the scale, videos are classified as 63-75: (excellent); as 51-62: (good); as 39-50: (moderate) as 28- 38: poor and as <28 very poor. JAMA scale is a scale that evaluates the accuracy, usefulness and reliability of video sources between 0-4. Insufficient data on 1 point video source obtained from the scale, refers to. 2-3 (partially sufficient data) and 4: (completely sufficient data)., Results: The evaluation on DISCERN scores given to the videos revealed that the average DISCERN score of the first researcher was 45.65±13.271 and the average of the second researcher's DISCERN score was 44.65±14.235. The average of JAMA scores was 2.67±0.826 for the first researcher and 2.49±0.845 for the second researcher. Accordingly, a statistically significant linear relationship was found between DISCERN (r=0.812) scores and JAMA (r=0.622) scores by the researchers., Conclusion: We found that YouTube has the potential to provide patients with easy access to a large amount of information on FGCS, although almost all of the videos we included in the study were uploaded by health professionals, the overall quality of the content on the subject was moderate. We strongly suggest that any videos on health should be uploaded by experts and these videos should definitely be reviewed by a referee., (Copyright © 2021. Published by Elsevier Masson SAS.)

In this paper, we aim to address issues of (1) joint spatial-temporal modeling and (2) side information injection for deep-learning based in-loop filter. For (1), we design a deep network with both progressive rethinking and collaborative learning mechanisms to improve quality of the reconstructed intra-frames and inter-frames, respectively. For intra coding, a Progressive Rethinking Network (PRN) is designed to simulate the human decision mechanism for effective spatial modeling. Our designed block introduces an additional inter-block connection to bypass a high-dimensional informative feature before the bottleneck module across blocks to review the complete past memorized experiences and rethinks progressively. For inter coding, the current reconstructed frame interacts with reference frames (peak quality frame and the nearest adjacent frame) collaboratively at the feature level. For (2), we extract both intra-frame and inter-frame side information for better context modeling. A coarse-to-fine partition map based on HEVC partition trees is built as the intra-frame side information. Furthermore, the warped features of the reference frames are offered as the inter-frame side information. Our PRN with intra-frame side information provides 9.0% BD-rate reduction on average compared to HEVC baseline under All-intra (AI) configuration. While under Low-Delay B (LDB), Low-Delay P (LDP) and Random Access (RA) configuration, our PRN with inter-frame side information provides 9.0%, 10.6% and 8.0% BD-rate reduction on average respectively. Our project webpage is https://dezhao-wang.github.io/PRN-v2/.

Idiopathic clubfoot is the most common congenital anomaly of the lower extremity. YouTube has emerged as an important source of health-related information for patients and families. Parents seek information about clubfoot on YouTube. However, the quality of these videos remains unknown. Therefore, we decided to evaluate the quality of YouTube videos about clubfoot. We searched YouTube for clubfoot videos using appropriate keywords. English language videos with more than 10 000 views were included. Three fellowship trained pediatric orthopedic surgeons independently assessed the videos and classified them into corporate, hospital, education and private. The quality of information was assessed using the Modified DISCERN and JAMA benchmark scores. The search yielded 12 060 videos of which 42 were analyzed. There were 9 (21%) videos from corporate organizations, 12 (29%) from hospitals, 3 (7%) from educational organizations and 18 (43%) by private individuals. The mean Modified DISCERN score was 2.1 ± 1.07 (range 0.3-4) and mean JAMA benchmark score was 0.9 ± 0.65 (range 0-2). Educational videos had highest mean Modified DISCERN score (3.1 ± 0.85) and private videos, the least (1.43 ± 1). This difference was statistically significant (P = 0.004). Hospital videos had highest mean JAMA benchmark score of 1.3 ± 0.6 as compared with private videos which had the least mean score of 0.5 ± 0.6. This difference was also statistically significant (P = 0.001). The results of our study indicate that the quality of information on idiopathic clubfoot on YouTube needs improvement. Videos from educational and hospital sources should be preferred over private sources.

This work presents a novel approach for the early recognition of the type of a laparoscopic surgery from its video. Early recognition algorithms can be beneficial to the development of "smart" OR systems that can provide automatic context-aware assistance, and also enable quick database indexing. The task is however ridden with challenges specific to videos belonging to the domain of laparoscopy, such as high visual similarity across surgeries and large variations in video durations. To capture the spatio-temporal dependencies in these videos, we choose as our model a combination of a convolutional neural network (CNN) and long short-term memory (LSTM) network. We then propose two complementary approaches for improving early recognition performance. The first approach is a CNN fine-tuning method that encourages surgeries to be distinguished based on the initial frames of laparoscopic videos. The second approach, referred to as " Future-State Predicting LSTM," trains an LSTM to predict information related to future frames, which helps in distinguishing between the different types of surgeries. We evaluate our approaches on a large dataset of 425 laparoscopic videos containing nine types of surgeries (Laparo425), and achieve on average an accuracy of 75% having observed only the first 10 min of a surgery. These results are quite promising from a practical standpoint and also encouraging for other types of image-guided surgeries.

Introduction: Social Media (SoMe) offers excellent opportunities for scientific knowledge dissemination and its use has been extended in urology. However, there is controversy about its use. Live videos shared trough SoMe platforms offer many advantages, but at the same time disadvantages and potential risks including confidentiality, copyright infringement, among others. We aimed to assess the activity of shared videos on SoMe during urological conferences., Materials and Methods: A comprehensive study of videos shared on SoMe during European Association of Urology congress was carried out from January 2016 to June 2018. The online tools Symplur (Symplur.com), Twitter, Periscope and YouTube were searched to collect data. Number of videos, transmission time and views were analyzed. Videos were classified as live or pre-recorded and as scientific or non-scientific. SPSS V22.0 was used to process data., Results: We identified 108 videos shared on SoMe, 292.42minutes of transmission, 67732 views. 79 of 108 (73%) were live streaming videos, 78 (72%) of which were considered scientific vs. 30 (28%) non-scientific. An increase was observed trough the years of study (2016-2018) in transmission time (p=.031) number of videos, views (p=.018) and live videos (p=.019) during the annual congress of the European Association of Urology., Conclusions: Shared videos on SoMe from urological conferences are increasing. These provide advantages for communication, scientific dissemination and expand the scope of conferences. However, there is potential risk of sharing information in real time; that could not be in line with the recommendations for appropriate use of social networks., (Copyright © 2019 AEU. Publicado por Elsevier España, S.L.U. All rights reserved.)

Objectives: We examined the content of YouTube(™) videos on mouth (oral) cancer and evaluated their usefulness in promoting early detection of oral cancer., Materials and Methods: A systematic search of YouTube(™) for videos containing information on mouth cancer was conducted using the keywords 'mouth cancer' and 'oral cancer'. Demographics of videos, including type, source, length, and viewers' interaction, were evaluated, and three researchers independently assessed the videos for usefulness in promoting early detection of oral cancer., Results: A total of 188 YouTube(™) videos (152 patient-oriented educational videos and 36 testimonial videos) were analyzed. The overall usefulness score ranged from 0 to 10 (mean = 3.56 ± 2.44). The most useful videos ranked late on the viewing list, and there was no significant correlation between video usefulness and viewing rate, viewers' interaction, and video length. Videos uploaded by individual users were less useful compared with videos uploaded by professional organizations or by healthcare professionals., Conclusion: Healthcare professionals, academic institutions, and professional organizations have a responsibility for improving the content of YouTube(™) about mouth cancer by uploading useful videos, and directing patients to reliable information sources., (© 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.)

Objectives: To quantitatively and qualitatively assess the characteristics of YouTube videos dealing with Sjögren's syndrome., Materials & Methods: A comprehensive electronic search was performed for 'Sjögren's syndrome' in YouTube. After excluding duplicates, irrelevant videos, and non-English-language domains, 70 videos were included for analysis. Videos were classified as useful, misleading, or personal experience. The overall quality of videos was scored according to the Global Quality Scale (GQS). Useful videos were assessed for reliability and comprehensiveness based on two 5-point scales. Key points of the misleading videos were explored and patients' personal experiences were further investigated., Results: Thirty-six videos (51.4%) were classified as useful, 6 (8.6%) as misleading, and 28 (40%) as patient's personal experience. Independent users tend to upload videos with personal experience, while university channels/professional organizations share useful videos with evidence-based information. Significant difference was observed in GQS among useful, misleading videos, and patient's experiences. The mean reliability, comprehensiveness, and GQS scores of useful videos were 2.5 (s.d.:1.2), 2.6 (s.d.:1.4), and 3.4 (s.d.:1.0), respectively, whereas only six videos (16.7%) were rated as complete vis-à-vis content. The most frequently misleading topics were etiology and treatment., Conclusions: Specialists should refer their patients to validated e-information resources and actively participate in the development of video-sharing platforms., (© 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.)

Despite important recent advances, the vulnerability of biometric systems to spoofing attacks is still an open problem. Spoof attacks occur when impostor users present synthetic biometric samples of a valid user to the biometric system seeking to deceive it. Considering the case of face biometrics, a spoofing attack consists in presenting a fake sample (e.g., photograph, digital video, or even a 3D mask) to the acquisition sensor with the facial information of a valid user. In this paper, we introduce a low cost and software-based method for detecting spoofing attempts in face recognition systems. Our hypothesis is that during acquisition, there will be inevitable artifacts left behind in the recaptured biometric samples allowing us to create a discriminative signature of the video generated by the biometric sensor. To characterize these artifacts, we extract time-spectral feature descriptors from the video, which can be understood as a low-level feature descriptor that gathers temporal and spectral information across the biometric sample and use the visual codebook concept to find mid-level feature descriptors computed from the low-level ones. Such descriptors are more robust for detecting several kinds of attacks than the low-level ones. The experimental results show the effectiveness of the proposed method for detecting different types of attacks in a variety of scenarios and data sets, including photos, videos, and 3D masks.

This paper presents a novel algorithm for terrain type classification based on monocular video captured from the viewpoint of human locomotion. A texture-based algorithm is developed to classify the path ahead into multiple groups that can be used to support terrain classification. Gait is taken into account in two ways. Firstly, for key frame selection, when regions with homogeneous texture characteristics are updated, the frequency variations of the textured surface are analyzed and used to adaptively define filter coefficients. Secondly, it is incorporated in the parameter estimation process where probabilities of path consistency are employed to improve terrain-type estimation. When tested with multiple classes that directly affect mobility-a hard surface, a soft surface, and an unwalkable area-our proposed method outperforms existing methods by up to 16%, and also provides improved robustness.

The goal of this article is to discuss the classification of video recordings and images when applied to dissolution testing in USP apparatus 1 and 2. Three use cases are presented. The first case presents the use and classification of video as RBE (review by exception) data under GAMP 5. The second case presents the use of video in formulation development in a research and development environment. The third case presents a feasibility study using readily available computer vision software to recognize and measure objects in the dissolution vessel, setting the groundwork for the use of image analysis as a quantitative tool. The classification of video as "electronic data", requiring 21 CFR part 11 compliance, versus its classification as a RBE data under GAMP 5, likely depends upon its use case. Another goal of this article is to establish a position on the use of video monitoring technology as a tool for dissolution testing that is fit for purpose and compliant with regulations regarding video data management and information.

We present a database of high-definition (HD) videos for the study of traits inferred from whole-body actions. Twenty-nine actors (19 female) were filmed performing different actions-walking, picking up a box, putting down a box, jumping, sitting down, and standing and acting-while conveying different traits, including four emotions (anger, fear, happiness, sadness), untrustworthiness, and neutral, where no specific trait was conveyed. For the actions conveying the four emotions and untrustworthiness, the actions were filmed multiple times, with the actor conveying the traits with different levels of intensity. In total, we made 2,783 action videos (in both two-dimensional and three-dimensional format), each lasting 7 s with a frame rate of 50 fps. All videos were filmed in a green-screen studio in order to isolate the action information from all contextual detail and to provide a flexible stimulus set for future use. In order to validate the traits conveyed by each action, we asked participants to rate each of the actions corresponding to the trait that the actor portrayed in the two-dimensional videos. To provide a useful database of stimuli of multiple actions conveying multiple traits, each video name contains information on the gender of the actor, the action executed, the trait conveyed, and the rating of its perceived intensity. All videos can be downloaded free at the following address: http://www-users.york.ac.uk/~neb506/databases.html. We discuss potential uses for the database in the analysis of the perception of whole-body actions.

Objective: To appraise a novel scoring system (Bedside Evaluation of Microcirculation [BEM]) to provide qualitative and semiquantitative assessment of canine microcirculatory videos generated by sidestream dark field imaging., Design: Prospective observational study., Setting: University teaching hospital., Animals: No animals were used in this study. Twenty microcirculatory videos (>20 s in length) acquired from the mucosa adjacent to the upper canine tooth of dogs were selected from a database of sidestream dark field microcirculatory videos with available current standard analysis (CSA)., Intervention: Three observers were trained to evaluate 5 video quality parameters (stability, content, illumination, focus, and pressure) and four perfusion parameters (total vessel density [TVD], capillary vessel density [CVD], perfused vessel density [PVD] and microvascular flow index [MFI]). Quality parameters were scored (excellent [0], sufficient [1], and insufficient [2]) similar to CSA recommendations. Each perfusion parameter was subjectively scored (1 lowest - 5 highest) using sample clips from the training video for comparison. Videos passed quality analysis if no parameter was scored insufficient. Repeatability and reproducibility were evaluated by assessing all films in a random order three times daily for 3 days. Strength of correlation of BEM with CSA for both qualitative and semiquantitative parameters was assessed., Measurements and Main Results: The qualitative evaluation pass/fail assessment matched CSA 86% of the time with individual observer agreements of 84-88%. Agreement with CSA did not change significantly over the study period (84%, 88%, and 84% on days 1, 2, and 3, respectively). No significant correlations were demonstrated between any BEM perfusion parameter and the corresponding CSA values., Conclusions: Rapid bedside assessment of microcirculatory video quality can be achieved. However, semiquantitative analysis by BEM demonstrated a lack of correlation with CSA for any of the perfusion parameters assessed., (© Veterinary Emergency and Critical Care Society 2014.)

We have developed a video processing method that achieves human perceptual visual quality-oriented video coding. The patterns of moving objects are modeled by considering the limited human capacity for spatial-temporal resolution and the visual sensory memory together, and an online moving pattern classifier is devised by using the Hedge algorithm. The moving pattern classifier is embedded in the existing visual saliency with the purpose of providing a human perceptual video quality saliency model. In order to apply the developed saliency model to video coding, the conventional foveation filtering method is extended. The proposed foveation filter can smooth and enhance the video signals locally, in conformance with the developed saliency model, without causing any artifacts. The performance evaluation results confirm that the proposed video processing method shows reliable improvements in the perceptual quality for various sequences and at various bandwidths, compared to existing saliency-based video coding methods.

Stimulus material for studying object-directed actions is needed in different research contexts, such as action observation, action memory, and imitation. Action items have been generated many times in individual laboratories across the world, but they are used in very few experiments. For future studies in the field, it would be worthwhile to have a larger set of action stimulus material available to a broader research community. Some smaller action databases have already been published, but those often focus on psycholinguistic parameters and static action stimuli. With this article, we introduce an action database with dynamic action stimuli. The database contains action descriptions of 1,754 object-directed actions that have been rated for familiarity in Germany and in China. For 784 of these actions, action video clips are available. With the use of our database, it is possible to identify actions that differ in familiarity between Western and Eastern cultures. This variable may be of interest to some researchers in the field, since it has been shown that familiarity influences action information processing. Action descriptions are listed and categorized in tables that can be downloaded, along with the corresponding video clips, as supplemental material.

Objective: This paper describes the investigation, categorization/characterization and viewing of pelvic floor muscle exercises (PFME) on YouTube from the perspective of the 'wisdom of the crowd'. The aim of the research was to increase awareness of the type of clips that individuals are likely to come across when searching YouTube and to describe trends and popularity. This awareness will be useful for the design of continence promotion services, especially for hard-to-reach individuals., Study Design: Web-based videos relating to PFE were identified by searching YouTube using the snowball technique. Main outcome measures Number of views; the approach taken (health, fitness, sexual and pregnancy); product promotion; and the use of music, visual cues and elements designed to encourage exercise. The number of views of each video was recorded at three points over a seven-month period., Results: Twenty-two videos were identified. Overall these videos had been viewed over 430,000 times during the study period. One video was viewed over 100,000 times and overall the median increase in views was 59.4%., Conclusions: YouTube is increasingly used to access information about pelvic floor exercises. Different approaches are used to communicate PFME information but there are no formal structures for quality control. Further research is required to identify which elements of the video clips are effective in communicating information and in motivating exercise and to establish appropriate protocols. Kitemarking is recommended in order that women obtain correct advice.

Purpose: This study is a preliminary examination of fire setting behaviors on YouTube., Methods: Data are based on a 1 day search of YouTube., Results: Results indicate YouTube is replete with inappropriate models of fire setting., Conclusion: The findings have potential implications for intervention with fire setting youth., (Copyright © 2012 Society for Adolescent Health and Medicine. Published by Elsevier Inc. All rights reserved.)

Digital video now plays an important role in medical education, health care, telemedicine and other medical applications. Several content-based video retrieval (CBVR) systems have been proposed in the past, but they still suffer from the following challenging problems: semantic gap, semantic video concept modeling, semantic video classification, and concept-oriented video database indexing and access. In this paper, we propose a novel framework to make some advances toward the final goal to solve these problems. Specifically, the framework includes: 1) a semantic-sensitive video content representation framework by using principal video shots to enhance the quality of features; 2) semantic video concept interpretation by using flexible mixture model to bridge the semantic gap; 3) a novel semantic video-classifier training framework by integrating feature selection, parameter estimation, and model selection seamlessly in a single algorithm; and 4) a concept-oriented video database organization technique through a certain domain-dependent concept hierarchy to enable semantic-sensitive video retrieval and browsing.

Digitalization of audio-visual resources combined with the performances of the networks offer many possibilities which are the subject of intensive work in the scientific and industrial sectors. Indexing such resources is a major challenge. Recently, the Motion Pictures Expert Group (MPEG) has been developing MPEG-7, a standard for describing multimedia content. The good of this standard is to develop a rich set of standardized tools to enable fast efficient retrieval from digital archives or filtering audiovisual broadcasts on the internet. How this kind of technologies could be used in the medical context? In this paper, we propose a simpler indexing system, based on Dublin Core standard and complaint to MPEG-7. We use MeSH and UMLS to introduce conceptual navigation. We also present a video-platform with enables to encode and give access to audio-visual resources in streaming mode.