Active Video Research Articles

INCREASING THE COMPETENCE OF TEACHERS OF SD KEPULAUAN SERIBU DKI JAKARTA PROVINCE THROUGH ASSISTANCE IN DEVELOPING PERFORMANCE ASSESSMENT INSTRUMENTS IN MATHEMATICS LEARNING

This community service activity will be carried out in order to carry out one of the strategic plans of the State University of Jakarta and the Faculty of Education, namely Educational Technology. This community service activity will be carried out in the Thousand Islands Administrative District, DKI Jakarta Province. Problems related to teacher competency development faced by elementary schools in Kelapa Island, Seribu Islands Administrative District, DKI Jakarta Province in implementing the 2013 Curriculum, one of which is about assessment in learning. Teachers were still not optimal in measuring assessment instruments to measure attitudes, knowledge, and skills. In general, teachers rarely designed the instruments used in assessing skills, but instead used the instruments contained in the teacher's book. One of the efforts to help solved the problems faced by elementary school teachers in Kelapa Island, Seribu Islands Administrative District, DKI Jakarta Province is to provide assistance with performance development instruments in learning Mathematics. The object (target audience) of those community service activity are elementary school teachers in Kelapa Island, Seribu Islands Administrative District, DKI Jakarta Province. The partners involved were the Head of the Seribu Islands Administrative District Education Office, DKI Jakarta Province and the Head of SDN 01 Kelapa Island, Seribu Islands Administrative District, DKI Jakarta. The outputs of community service activities in the form of products/goods are: (a) performance assessment instruments in elementary mathematics learning; (b) publication of activity videos on Youtube; (c) partner's certificate regarding an increase in partner empowerment regarding teacher competence in developing performance assessment instruments in elementary mathematics learning; and (d) nameplate proof of the Assisted Village Cooperation. The service output is assistance in developing assessment instruments in elementary mathematics learning. Other outputs are articles in ISSN journals/ISSN national service seminars

Video-based evaluation system for tic action in Tourette syndrome: modeling, detection, and evaluation.

Behavioral ratings based on clinical observations are still the gold standard for screening, diagnosing, and assessing outcomes in Tourette syndrome. Detecting tic symptoms plays an important role in patient treatment and evaluation; accurate tic identification is the key to clinical diagnosis and evaluation. In this study, we proposed a tic action detection method using face video feature recognition for tic and control groups. Through facial ROI extraction, a 3D convolutional neural network was used to learn video feature representations, and multi-instance learning anomaly detection strategy was integrated to construct the tic action analysis and discrimination framework. We applied this tic recognition framework in our video dataset. The model evaluation results achieved average tic detection accuracy of 91.02%, precision of 77.07% and recall of 78.78%. And the tic score curve with postprocessing provided information of how the patient's twitches change over time. The detection results at the individual level indicated that our method can effectively detect tic actions in videos of Tourette patients without the need for fine labeling, which is significant for the long-term evaluation of patients with Tourette syndrome.

Active video game (AVG); a viable option for the promotion of physical activity.

Active video game (AVG); a viable option for the promotion of physical activity.

Attentional Composition Networks for Long-Tailed Human Action Recognition

The problem of long-tailed visual recognition has been receiving increasing research attention. However, the long-tailed distribution problem remains underexplored for video-based visual recognition. To address this issue, in this article we propose a compositional learning based solution for video-based human action recognition. Our method, named Attentional Composition Networks (ACN), first learns verb-like and preposition-like components, then shuffles these components to generate samples for the tail classes in the feature space to augment the data for the tail classes. Specifically, during training, we represent each action video by a graph that captures the spatial-temporal relations (edges) among detected human/object instances (nodes). Then, ACN utilizes the position information to decompose each action into a set of verb and preposition representations using the edge features in the graph. After that, the verb and preposition features from different videos are combined via an attention structure to synthesize feature representations for tail classes. This way, we can enrich the data for the tail classes and consequently improve the action recognition for these classes. To evaluate the compositional human action recognition, we further contribute a new human action recognition dataset, namely NEU-Interaction (NEU-I). Experimental results on both Something-Something V2 and the proposed NEU-I demonstrate the effectiveness of the proposed method for long-tailed, few-shot, and zero-shot problems in human action recognition. Source code and the NEU-I dataset are available at https://github.com/YajieW99/ACN .

Cross-modal alignment and translation for missing modality action recognition

Multimodal data provide complementary information on the same context, leading to performance improvement in video action recognition. However, in reality, not all modalities are available at test time. To this end, we propose Cross-Modal Alignment and Translation (CMAT) framework for action recognition that is robust to missing modalities. Specifically, our framework first aligns representations of multiple modalities from the same video sample through contrastive learning, effectively alleviating the bias with respect to the type of missing modality. Then, CMAT learns to translate representations of one modality into that of another modality. This allows the representations of the missing modalities to be generated from the remaining modalities during the testing. Accordingly, CMAT fully utilizes multimodal information obtained through abundant interactions across modalities. The proposed CMAT achieves state-of-the-art performances in both complete and missing modality settings on NTU RGB+D, NTU RGB+D 120, and Northwestern-UCLA datasets. Moreover, extensive ablation studies demonstrate the effectiveness of our design.

A heterogeneous two-stream network for human action recognition

The most widely used two-stream architectures and building blocks for human action recognition in videos generally consist of 2D or 3D convolution neural networks. 3D convolution can abstract motion messages between video frames, which is essential for video classification. 3D convolution neural networks usually obtain good performance compared with 2D cases, however it also increases computational cost. In this paper, we propose a heterogeneous two-stream architecture which incorporates two convolutional networks. One uses a mixed convolution network (MCN), which combines some 3D convolutions in the middle of 2D convolutions to train RGB frames, another one adopts BN-Inception network to train Optical Flow frames. Considering the redundancy of neighborhood video frames, we adopt a sparse sampling strategy to decrease the computational cost. Our architecture is trained and evaluated on the standard video actions benchmarks of HMDB51 and UCF101. Experimental results show our approach obtains the state-of-the-art performance on the datasets of HMDB51 (73.04%) and UCF101 (95.27%).

Dynamic Context Removal: A General Training Strategy for Robust Models on Video Action Predictive Tasks

AbstractPredicting future actions is an essential feature of intelligent systems and embodied AI. However, compared to the traditional recognition tasks, the uncertainty of the future and the reasoning ability requirement make prediction tasks very challenging and far beyond solved. In this field, previous methods usually care more about the model architecture design but little attention has been put on how to train models with a proper learning policy. To this end, in this work, we propose a simple but effective training strategy, Dynamic Context Removal (DCR), which dynamically schedules the visibility of context in different training stages. It follows the human-like curriculum learning process, i.e., gradually removing the event context to increase the prediction difficulty till satisfying the final prediction target. Besides, we explore how to train robust models that give consistent predictions at different levels of observable context. Our learning scheme is plug-and-play and easy to integrate widely-used reasoning models including Transformer and LSTM, with advantages in both effectiveness and efficiency. We study two action prediction problems, i.e., Video Action Anticipation and Early Action Recognition. In extensive experiments, our method achieves state-of-the-art results on several widely-used benchmarks.

Interventions targeting children and young people's physical activity behavior at home: A systematic review.

Increased time at home during the COVID-19 pandemic significantly decreased children's physical activity. This systematic review aimed to evaluate the effectiveness of children's home-based physical activity interventions, and identify 'active ingredients' underpinning these. Databases searched-AMED, PsychINFO, CINAHL, Cochrane, EMBASE, PubMed/Medline, Scopus, SPORTDiscus and Web of Science, from inception until June 2022. Eligibility criteria-children aged 2-16 years, targeting home-based physical activity, a control group, and physical activity measured pre- and post- intervention. Studies were excluded if it was not possible to identify change in physical activity at home. The review was written following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance. Study quality was evaluated using the quality assessment tool for quantitative studies. Study design, intervention characteristics, outcome data, behavior change theory, Behavior Change Techniques (BCTs) and process evaluation data were extracted and discussed using narrative syntheses. 13 studies (including 1,182 participants) from 25,967 were included. Interventions primarily involved active video games, with the addition of coaching or telehealth support (n = 5). Three of the 13 studies significantly increased children's physical activity (1 = Moderate to vigorous physical activity, 2 = total volume, P<0.05). The largest effect size (d = 3.45) was for moderate to vigorous physical activity. 29% of BCTs were identified across included interventions; the most common being adding objects to the environment. The most effective intervention scored strong for design quality, incorporated telehealth coaching, and included the most commonly coded BCTs. Variation among studies and insufficient reporting of data made a meta-analysis unfeasible. COVID-19 emphasized the importance of the home for physical activity. Whilst effectiveness of interventions was limited, building social support and self-efficacy are mechanisms that should be explored further. The review provides recommendations to improve the design and evaluation of future interventions. Prospero registration number: CRD42020193110.

The Usability of a Touchpad Active Video Game Controller for Individuals With Impaired Mobility: Observational Study.

Video games are a popular sedentary activity among people with impaired mobility; however, active video game hardware typically lacks accessibility and customization options for individuals with mobility impairments. A touchpad video game system can elicit moderate physical activity in healthy adults; however, it is unclear if this system is usable by adults with impaired mobility. The purpose of this study was to assess the usability of a touchpad video game controller system adapted for adults with impaired mobility. Additional outcomes explored were enjoyment, perceived exertion, self-efficacy, participant feedback, and researcher observations of gameplay. Participants played several video game titles for 20 minutes with a touchpad video game controller as they stood or sat in a chair or their wheelchair. Usability was assessed with the System Usability Scale (SUS) and the Health Information Technology Usability Evaluation Scale (Health-ITUES) surveys after gameplay. After each video game, participants reported enjoyment using a visual analog scale (0 to 100 mm) and a rating of perceived exertion using the OMNI 0 to 10 scale. Self-efficacy was measured before and after gameplay. Participants provided feedback at the end of their session. In total, 21 adults (6 females and 15 males) with a mean age of 48.8 (SD 13.8) years with various mobility impairments participated in this study. The touchpads received mean usability scores on the SUS 80.1 (SD 18.5) and Health-ITUES 4.23 (SD 0.67). The SUS scores reported suggest the touchpad system is "usable"; however, the Health-ITUES scores were slightly below a suggested benchmark. Participants reported moderate to high enjoyment but perceived the exertion as "somewhat easy." Self-efficacy was moderate to high and did not differ pre- to postgame play. The participants regarded the touchpads as novel, fun, and entertaining. The generalizability of our results is limited due to the heterogenous sample; however, our participants identified several areas of improvement for future iteration.

The relationship between the brain-derived neurotrophic factor and neurocognitive response to physical exercise in individuals with schizophrenia

ObjectivePhysical exercise can improve neurocognition in individuals with schizophrenia, presumably by facilitating neuroplasticity. There is, however, large inter-individual variation in response. The brain-derived neurotrophic factor (BDNF) has been proposed to mediate these effects. The current aim was to investigate the sparsely studied relationship between peripheral resting BDNF and neurocognitive response to physical exercise in individuals with schizophrenia. MethodThe current study reports secondary analyses of data from a randomized controlled trial (RCT), ClinicalTrials.gov number 02205684, recently reported according to the CONSORT guidelines. Eighty-two individuals with schizophrenia (mean age 37 ± 14 years old, 61% men) were randomly allocated to high-intensity interval training (HIIT) or a comparison group performing low-intensity active video gaming (AVG). Both interventions consisted of 2 sessions/week for 12 weeks. In previously published primary RCT analyses, HIIT and AVG showed comparable small to moderate improvements in neurocognition. We now address the inter-individual variability in neurocognitive response. We apply mediation and moderation analyses for repeated measures designs (MEMORE) and mixed effects models. ResultsBaseline neurocognition was not significantly correlated with baseline levels of mature BDNF (baseline-mBDNF) or the precursor proBDNF. Nonetheless, baseline-mBDNF, but not baseline proBDNF, moderated the effect of exercise on neurocognition (p = 0.025) and explained 7% of the variance. The neurocognitive improvement increased with increasing baseline-mBDNF values. The moderating effect of baseline-mBDNF remained significant in a more complex model adding the moderating effects of exercise mode, sex, age, duration of illness and baseline VO2max on the outcome (neurocognition). Mean baseline-mBDNF significantly decreased from baseline to post-intervention (p = 0.036), regardless of exercise mode, differing by sex and associated with improved VO2max but not with change in neurocognition. A mediating role of mBDNF on the effect of physical exercise on neurocognition was not supported. Values of proBDNF mainly remained stable from baseline to post-intervention. ConclusionWe found that baseline-mBDNF moderated the effect of physical exercise on neurocognition in individuals with schizophrenia and explained a small part of the inter-individual variation in neurocognitive response. Mean mBDNF decreased from baseline to post-intervention, regardless of exercise mode. A mediating role of mBDNF on the effect of exercise on neurocognition was not supported. The inter-individual variation in neurocognitive response and the complex role of peripheral BDNF in physical exercise is still to be elucidated.

Video Action Recognition by Combining Spatial-Temporal Cues with Graph Convolutional Networks

Video action recognition relies heavily on the way spatio-temporal cues are combined in order to enhance recognition accuracy. This issue can be addressed with explicit modeling of interactions among objects within or between videos, such as the graph neural network, which has been shown to accurately model and represent complicated spatial- temporal object relations for video action classification. However, the visual objects in the video are diversified, whereas the nodes in the graphs are fixed. This may result in information overload or loss if the visual objects are too redundant or insufficient for graph construction. Segment level graph convolutional networks (SLGCNs) are proposed as a method for recognizing actions in videos. The SLGCN consists of a segment-level spatial graph and a segment-level temporal graph, both of which are capable of simultaneously processing spatial and temporal information. Specifically, the segment-level spatial graph and the segment-level temporal graph are constructed using 2D and 3D CNNs to extract appearance and motion features from video segments. Graph convolutions are applied in order to obtain informative segment-level spatial-temporal features. A variety of challenging video datasets, such as EPIC-Kitchens, FCVID, HMDB51 and UCF101, are used to evaluate our method. In experiments, it is demonstrated that the SLGCN can achieve performance comparable to the state-of-the-art models in terms of obtaining spatial-temporal features.

Higher working memory capacity and distraction-resistance associated with strategy (not action) game playing in younger adults, but puzzle game playing in older adults

Superior attention and Working Memory (WM) have been reported for habitual action video gamers compared to other gamers or non-players. With an online experiment we measured visuo-spatial WM capacity and ability to ignore distraction, and participants listed the video games they played. Categorising the 209 young adult participants (18–30 years) according to the game type they predominantly played revealed superior WM capacity for strategy and action gamers compared to non-players. However, re-categorising the games according to their constituent game types revealed superior WM capacity and distraction resistance associated with strategy but not action game components. In contrast to younger adults, data from 181 older adults (60–81 years) showed superior WM capacity and distractor-resistance for puzzle gamers, which was equivalent to that of younger adults. The results highlight the need to consider component game types in games research and inform the design of age-appropriate cognitive interventions.

Mutual modality learning for video action classification

The construction of models for video action classification progresses rapidly. However, the performance of those models can still be easily improved by ensembling with the same models trained on different modalities (e.g. Optical flow). Unfortunately, it is computationally expensive to use several modalities during inference. Recent works examine the ways to integrate advantages of multi-modality into a single RGB-model. Yet, there is still room for improvement. In this paper, we explore various methods to embed the ensemble power into a single model. We show that proper initialization, as well as mutual modality learning, enhances single-modality models. As a result, we achieve state-of-the-art results in the Something-Something-v2 benchmark.

AGPN: Action Granularity Pyramid Network for Video Action Recognition

Video action recognition is a fundamental task for video understanding. Action recognition in complex spatio-temporal contexts generally requires fusing of different multi-granularity action information. However, existing works do not consider spatio-temporal information modeling and fusion from the perspective of action granularity. To address this problem, this paper proposes an Action Granularity Pyramid Network (AGPN) for action recognition, which can be flexibly integrated into 2D backbone networks. The core module is the Action Granularity Pyramid Module (AGPM), a hierarchical pyramid structure with residual connections, which is established to fuse multi-granularity action spatio-temporal information. From top to bottom level in the designed pyramid structure, the receptive field decreases and action granularity becomes more refined. To enrich temporal information of the inputs, a Multiple Frame Rate Module (MFM) is proposed to mix different frame rates at a fine-grained pixel-wise level. Moreover, a Spatio-temporal Anchor Module (SAM) is employed to fix spatio-temporal feature anchors to promote the effectiveness of feature extraction. We conduct extensive experiments on three large-scale action recognition datasets, Something-Something V1 & V2 and Kinetics-400. The results demonstrate that our proposed AGPN outperforms the state-of-the-art methods for the tasks of video action recognition.

Perceiving local relative motion and global correlations for weakly supervised group activity recognition

This paper presents a weakly supervised approach for group activity recognition by exploiting the local relative motion and global correlations among entities. Most existing approaches of group recognition characterize motions of group activities based on the coordinates of individuals or feature maps without excluding the camera motion, which is a combination of local relative motion and camera motion. To address this problem, we utilize a simple but effective Local Relative Motion Module (LRMM): a 3D-CNN-based network to exploit the local movement. We further employ a Global Correlation Module (GCM) to establish relationships among different feature patches for capturing the entire scene. We have evaluated the proposed method on sports and group activity video. The method has achieved state-of-the-art performance on three challenging datasets for weakly supervised group activity recognition. The method has also outperformed some approaches trained with much stronger supervision in the comparative evaluation.

Action Recognition using Spatial and Temporal Features with Kernel SVM

A new low-level visual feature, called Spatio-temporal context distribution feature of interest points is used to describe human actions. Each action video is expressed as a set of relative XYT coordinates between interest points listed pair wise in a local region. From the input image frames the Locally Weighted Word Context (LWWC ) descriptor encodes the spatial context interest points rather than being limited to a single interest point and the Graph Regularized Nonnegative Matrix Factorization (GNMF) is used to encode the geometrical information by constructing a nearest neighbour graph. By extracting the kernel weights of the obtained feature variables , the kernel weighted SVM is modelled to jointly capture the compatibility between multilevel action features and action classes and the compatibility between multilevel scene features and scene classes. The contextual relationship between action classes and scene classes is derived using the kernel weight as a variable.

Increasing Community Awareness of Natural Disasters in the Context of Disaster Mitigation in Kalinongko Hamlet, Pagerharjo, Samigaluh, Kulon Progo

Kalinongko Hamlet is a hamlet located in Pagerharjo, Kulon Progo, Special Region of Yogyakarta. Kemesu Hamlet has a geographical location in the Pagerharjo area. Pagerharjo Village has boundaries with Magelang Regency in the north, Ngargosari Village in the south, Banjarsari Village in the east, and Purworejo Regency in the west. Kulon Progo Regency including Pagerharjo Village, especially Kemesu Hamlet, is one of the districts that has the potential for disasters such as floods, tsunamis, earthquakes, landslides, droughts, and volcanic eruptions. Based on data obtained from BNPB, the Pagerharjo area has a high potential for landslides and drought. On the other hand, also based on data obtained from BNPB, the capacity to deal with landslides was in the medium and low categories in dealing with drought hazards. Therefore, efforts need to be made to increase the capacity of the community in Pagerharjo Village, especially in Kalinongko Hamlet, in dealing with potential landslides and drought disasters. This PKM KKN PPM activity seeks to increase community capacity in dealing with disasters through socialization and Discussion Group Forums (FGD). Resource persons for this activity are planned from disaster actors either from the campus or from the Regional Disaster Management Agency (BPPD). With this disaster socialization and FGD activity, it is hoped that it can increase the community's capacity to deal with potential disasters. The output of this PKM is published in the mass media, seminar forums, and activity videos.

Skill Enhancement Through Electronics Installation and Maintenance Training For Forum Jamaah Mushola Al-Hidayah Members

At present, electronic equipment is one of the most important means for the smooth implementation of worship. One of them is a sound system, which is very commonly used during events such as recitations, sermons, and other religious activities. Over time, the influence of environmental factors, usage age, poor installation, and poor maintenance can cause the quality of electronic devices to decrease and even the components to deteriorate quickly. This is generally due to improper installation, poor operation, or poor maintenance. This service is planned to partner with the Jamaah Forum and Takmir Musala Al Hidayah in Kragilan Hamlet, Sidoluhur, Godean, where there is a problem with the lack of an optimal indoor sound system in the prayer room. The solutions offered to overcome these problems are development, installation, and operational training, as well as maintenance of indoor prayer sound systems and procurement of portable sound systems. The output targets of this program are publications in scientific forums and mass media, videos of community service activities, and the transfer of technology applications to partners. In order for the service objectives to be achieved, the service team uses several stages, namely the initial stage in the form of analyzing the needs of partners by discussing with them, preparing service proposals, the program implementation stage, making publication outputs and final reports, and the evaluation stage of service activities. All of the series of service activities have been carried out according to plan and are running well. Members of the Al Hidayah Jamaah forum participated in the activity, which was training for electronics installation and maintenance. At the end of the activity, a partner satisfaction survey was conducted with satisfactory results.

Cross-domain video action recognition via adaptive gradual learning

Video-based Unsupervised Domain Adaptation (UDA) methods concentrate on addressing domain shift and improving the robustness of video models. It can be naturally applied to cross-domain action recognition tasks, but the inherent complexity of videos makes this task more challenging. Though several recent attempts have achieved superior performance in the video UDA field, most of them intend to mitigate domain shift directly from the source domain to the target domain. These methods will cause a performance drop when the domain distribution shift is large. To better realize the domain adaptation, this paper proposes an effective gradual learning framework by constructing multiple auxiliary domains to achieve progressive transfer for cross-domain video action recognition. Specifically, a Dynamic CutMix Mechanism (DCM) is introduced to build the auxiliary domains that mitigate the domain gap caused by object distance and background discrepancies. Furthermore, the Gradual Transfer Strategy (GTS) utilizes these auxiliary domains to realize the cross-domain action classification gradually. Extensive experiments validate the effectiveness of our proposed method, and the experimental results can significantly outperform state-of-the-art methods on multiple benchmark cross-domain datasets.

The Use of Educative Game Tools (Meronce Lego) to Support the Motor Development of Preschool Children at TA Pesan Ibu

Pre-school period is the stage of development of children aged 3-6 years where at this age children will learn to interact with other people. Educational Game Tools (APE) are all forms of games designed to provide educational experiences or learning experiences to the players. Message Mother Kindergarten is one of the places to stimulate the development of preschool-aged children, however APE in the form of Meronce Lego available at schools is still not optimal where the availability of educational game tools itself does not match the number of children so there is a need for additions and understanding to improve children's development and Children feel bored with monotonous games. The solution that will be provided is to provide training in the Selection of Educational Game Tools (APE), Assistance explaining the uses and functions of Educational Game Tools (APE) and Promotion of Educational Game Tools (APE). The output target to be produced is in the form of activity videos which are published via YouTube with the link: https://youtu.be/ZISk0JYROWE. All students can play Meronce Lego happily with their friends. Thus the TA teacher, especially the TA Pesan Ibu, can introduce educational game tools such as lego meronce which is appropriate for one of the lessons in the classroom so that children don't feel bored with this type of game. Students play the lego meronce game so that it can improve children's motor development.