Analysis Of Video Data Research Articles

A Non-Invasive Gait-Based Screening Approach for Parkinson’s Disease

Parkinson's disease (PD) presents a significant global health challenge, characterized by the progressive degeneration of dopamine-producing neurons in the brain, resulting in both motor and non-motor symptoms that severely impact quality of life. This study addresses the complexities of PD, highlighting the critical need for early diagnosis to slow disease progression. This research addresses the challenges of early diagnosis, such as the use of unreliable diagnostic techniques and limited healthcare resources. It uses the MMU Parkinson Disease Dataset and applies camera-based data collection to analyze gait patterns that can identify a risk of Parkinson's Disease. The study utilizes computer vision and the AlphaPose framework to analyze video data and detect body key points. By employing machine learning algorithms, including Support Vector Machines (SVM) and CatBoost, showing highly effective in identifying temporal dependencies in gait patterns. The algorithms achieved a high accuracy of 83.33% on the MMU dataset. This method enhances the accuracy of PD detection and enables immediate detection and control of the disease. The combination of advanced data analysis methods and medical knowledge offers new possibilities to develop targeted treatments that improve patient outcomes, demonstrating the potential of machine learning in effectively managing and treating Parkinson's disease. To enhance the generalizability of models, future research should collect extensive and diverse datasets covering various backgrounds and different stages of Parkinson's disease and utilize advanced techniques for extracting features to improve the accuracy of gait analysis.

Yo! Chisme Is Good! Latina/x Girls’ Cultivando Confianza in an After-School Program

Background or Context: Middle-school Girls of Color experience complex systems of relationships, structures, and situations in and out of school spaces. Their experiences, however, are often at the margins or excluded from middle school curricula and after-school programming. In this article, we focus on how young women engage with chisme to liberate themselves from social and academic pressures placed upon them. Purpose and Research Questions: Our study seeks to center and explore the ways of knowing and being of Latina/x middle school girls with particular attention to the role of chisme. We draw on Chicana/Latina feminist perspectives to address the following research questions: (1) How do middle school Latina/x girls define chisme? (2) How and why was chisme enacted or leveraged to co-create a trusting space in an after-school program for middle school Latina/x girls? Research Design: We employed a practitioner research inquiry design to explore chisme as a literacy practice of Latina/x middle-school girls in an after-school program. The after-school program, Latina/x Girlhood Narratives, was implemented in a middle school in a mid-sized, urban city in the Midwest region of the United States. Analytically, we engaged in video data analysis and qualitative coding analysis to examine the data. Results and Conclusions: Three main findings emerged from the study: (1) shared understandings: chismes, rumors, and snitching; (2) todo or la mitad del chisme: a process and product of confianza; and (3) from chismes to liberatory literacy practices of desahogamiento. Our study demonstrates how chisme, in particular, not only was vital to building and deepening confianza, but also allowed access to the girls’ ideas for activities and pedagogies. The field of education must continue to work to make visible the ways in which Latina/x girls engage with themselves and the world in order to sustain better relationships with and for them, and to build pathways for Latina/x girls to see themselves as successful in schools and in society.

Automated system for classifying images of video streams for timely detection of fires

ABSTRACT Early detection of a fire during its development stage is highly desired to minimize human and material losses. The improvement of the efficiency of these monitoring systems through the analysis of fire video data drawn from unmanned aerial vehicles is considered . Classification of video images when monitoring a fire event on the territory is used based on their segmentation into identical rectangular segments of a given size and then assigning them to one of three classes: Smoke, Flame, and Indifferent classes. The Walsh-Hadamard transform was used to form descriptors for the Weak classifiers. Descriptors were calculated for three Weak classifiers; first, the Walsh-Hadamard transform was calculated for the window of the entire segment, and its spectral coefficients were used for the first Weak classifier. Next, the descriptors were calculated in windows whose sizes were half and a quarter of the original window size. The classifier consisted of three independently-trained neural networks. A simple ensemble averaging block was used to combine the outputs of those networks. A special software code has been developed to classify video images. The code allows us to create a database of segments of the ‘Smoke’ and ‘Flame’ classes, determine the two-dimensional Walsh-Hadamard spectrum of video image segments, train fully connected neural networks, conduct exploratory analysis and evaluate the relevance of the calculated two-dimensional spectral coefficients. For validation we used real data from Сlosed Circuit Television cameras presented in open databases. The experimental studies on the classification of video data containing flames and smoke showed that the proposed method has an average smoke detection accuracy of 86% and an average flame detection accuracy of 89.5%. Errors II in smoke detection and flame detection had the averages of 13% and 4.5%, respectively.

Automated waste detection system for river surveillance in Jakarta using background substraction method

Abstract In Jakarta, the issue of river pollution due to indiscriminate waste disposal poses serious environmental and safety concerns, often leading to flooding during the rainy season. Manual surveillance by human resources has proven ineffective in addressing the escalating scale of the problem. This study presents an automated waste detection system for river surveillance in Jakarta, especially inorganic waste, utilizing video processing techniques, specifically background subtraction and frame differencing. We collected and analyzed video data from 13 rivers, including the Ciliwung, Angke, and Pesanggrahan rivers, during October and November 2023. The system’s performance was evaluated based on its accuracy in detecting waste objects, with detection rates varying significantly across different rivers. High detection accuracies were achieved in the Cipinang (93%) and Malang (90%) rivers, while lower accuracies were noted in rivers like Grogol (25%) and Cakung (17%). The overall average detection rate was 60%. These results highlight the system’s strengths in cleaner, less dynamic environments and its challenges in more complex conditions. Future research should focus on enhancing algorithm robustness, incorporating adaptive thresholding, and integrating multi-sensor data to improve detection accuracy.

Object Detection and Classification Framework for Analysis of Video Data Acquired from Indian Roads

Object detection and classification in autonomous vehicles are crucial for ensuring safe and efficient navigation through complex environments. This paper addresses the need for robust detection and classification algorithms tailored specifically for Indian roads, which present unique challenges such as diverse traffic patterns, erratic driving behaviors, and varied weather conditions. Despite significant progress in object detection and classification for autonomous vehicles, existing methods often struggle to generalize effectively to the conditions encountered on Indian roads. This paper proposes a novel approach utilizing the YOLOv8 deep learning model, designed to be lightweight, scalable, and efficient for real-time implementation using onboard cameras. Experimental evaluations were conducted using real-life scenarios encompassing diverse weather and traffic conditions. Videos captured in various environments were utilized to assess the model’s performance, with particular emphasis on its accuracy and precision across 35 distinct object classes. The experiments demonstrate a precision of 0.65 for the detection of multiple classes, indicating the model’s efficacy in handling a wide range of objects. Moreover, real-time testing revealed an average accuracy exceeding 70% across all scenarios, with a peak accuracy of 95% achieved in optimal conditions. The parameters considered in the evaluation process encompassed not only traditional metrics but also factors pertinent to Indian road conditions, such as low lighting, occlusions, and unpredictable traffic patterns. The proposed method exhibits superiority over existing approaches by offering a balanced trade-off between model complexity and performance. By leveraging the YOLOv8 architecture, this solution achieved high accuracy while minimizing computational resources, making it well suited for deployment in autonomous vehicles operating on Indian roads.

Managing the rehabilitation of patients following joint replacement using remote support tools

The article considers the approach to the development of an intelligent decision support system based on artificial intelligence technologies to manage the rehabilitation process in patients who have undergone joint endoprosthetic surgery. The paper focuses on the advantages of using complex intelligent systems in various industries requiring personalised approaches to the object, including medicine. The prospects for their further development are considered. The creation of a digital twin reflecting the key parameters of the object and its development trajectory from the moment of the first interaction with the system is proposed. The system, developed on the basis of software and instrumental environments, provides the possibility of remote observation and correction of the recovery plan by specialists, which represents a significant advantage compared to traditional methods. In order to ensure the adaptability of the system and the actualisation of the proposed solutions, control over the patient's condition and adaptation of treatment programmes is achieved by means of periodic and unscheduled checking of the object indicators. This is achieved by comparing the indicators obtained as a result of video data analysis with the normalised indicators of the system. The developed system analyses video recordings, performs movement analysis and adapts rehabilitation measures depending on the obtained data on the patient's motor activity in real time. The proposed method offers the potential to enhance the precision and personalisation of rehabilitation approaches, thereby facilitating greater accessibility and efficacy in treatment. The integration of artificial intelligence elements into clinical practice presents a promising avenue for optimising rehabilitation processes.

Smartphone Assessment of the Sitting Heel-Rise Test.

The study presents a new approach for assessing plantarflexor muscles' function using a smartphone. The test involves performing repeated heel raises for 60 s while seated. The seated heel-rise test offers a simple method for assessing plantarflexor muscles' function in those with severe balance impairment who are unable to complete tests performed while standing. The study aimed to showcase how gyroscopic data from a smartphone placed on the lower limb can be used to assess the test. Eight participants performed the seated heel-rise test with each limb. Gyroscope and 2D video analysis data (60 Hz) of limb motion were used to determine the number of cycles, the average rise (T-rise), lowering (T-lower), and cycle (T-total) times. The number of cycles detected matched exactly when the gyroscope and kinematic data were compared. There was good time domain agreement between gyroscopic and video data (T-rise = 0.0005 s, T-lower = 0.0013 s, and T-total = 0.0017 s). The 95% CI limits of agreement were small (T-total -0.1118, 0.1127 s, T-lower -0.1152, 0.1179 s, and T-total -0.0763, 0.0797 s). Results indicate that a smartphone placed on the thigh can successfully assess the seated heel-rise test. The seated heel-rise test offers an attractive alternative to test plantarflexor muscles' functionality in those unable to perform tests in standing positions.

Co-Construction of Iteration Through Failure Moments: Interactions Between Museum Educators and Visitors

Museum educators play a major role in how visitors’ experience failure moments during STEM-related activities. The purpose of this study was to explore how museum educators co-constructed iteration through failure moments with visitors during an engineering activity. Utilizing an instrumental case study, we analyzed video data and one-on-one reflective meetings from five museum educators. Through our analysis, we highlight how educators and visitors are able to jointly attend, interpret, and respond to failures that leads to continuous improvements of the prototype and/or design process (i.e., iteration). The significance of this study lies in providing informal educators with approaches they can incorporate to support visitors during the failure-learning process, namely, strategies that develop visitors’ noticing skills around failure.

Sensing emotional valence and arousal dynamics through automated facial action unit analysis

Information about the concordance between dynamic emotional experiences and objective signals is practically useful. Previous studies have shown that valence dynamics can be estimated by recording electrical activity from the muscles in the brows and cheeks. However, whether facial actions based on video data and analyzed without electrodes can be used for sensing emotion dynamics remains unknown. We investigated this issue by recording video of participants’ faces and obtaining dynamic valence and arousal ratings while they observed emotional films. Action units (AUs) 04 (i.e., brow lowering) and 12 (i.e., lip-corner pulling), detected through an automated analysis of the video data, were negatively and positively correlated with dynamic ratings of subjective valence, respectively. Several other AUs were also correlated with dynamic valence or arousal ratings. Random forest regression modeling, interpreted using the SHapley Additive exPlanation tool, revealed non-linear associations between the AUs and dynamic ratings of valence or arousal. These results suggest that an automated analysis of facial expression video data can be used to estimate dynamic emotional states, which could be applied in various fields including mental health diagnosis, security monitoring, and education.

Turnstile politics: practices of care and mobility justice in Santiago’s public transport system

Modern public transport systems are typically designed by following universal aspirations to predictability and standardisation. In the case of Santiago’s public transport, however, this design philosophy has often translated into concrete, practical struggles for users with more vulnerable corporealities. In analysing the case of a controversial turnstile installed in Santiago’s buses in 2016, this paper draws on video analysis to examine how passengers respond to, and locally deal with, its exclusionary design. Passengers’ interactions with this technology demonstrate how the turnstile is taken up as more than a mere sorting device, to become a matter of concern around mobility (in)justice. A detailed analysis of these interactions – through ethnomethodological analysis of video data – describes how passengers respond to the turnstile’s exclusionary design by deploying diverse embodied practices of care towards other users experiencing trouble with it. Such practices of care align with a grounded, embodied sense of mobility (in)justice as opposed to more abstract understandings of distributive justice. The paper concludes by discussing the potential forpublic transport systems to become more just and inclusive environments designed as infrastructures of care, whereby materialities may not only allow, but support, practices of care among users.

EXCOR membrane motion analyzer (EMMA) to quantify and assess hemodynamic performance of the EXCOR pediatric heart assist device.

Pediatric heart failure is associated with high mortality rates and is a current clinical burden. There is only one FDA approved pediatric VAD, Berlin Heart EXCOR, for treatment. Thrombo-embolic complications are a significant clinical challenge, which can lead to devastating complications such as stroke and impair efficient EXCOR function. Currently, clinicians perform largely qualitative periodic assessment of EXCOR operation by observing the motion of a rapidly moving membrane, which can be prone to human error and can lead to missing out on crucial information. In this study, we design and implement a quantitative early warning system for accurate and quantitative assessment of the EXCOR membrane, named EXCOR Membrane Motion Analyzer (EMMA). Using a combination of image analysis, computer vision and custom designed algorithm, we perform rigorous frame by frame analysis of EXCOR membrane video data. We developed specialized metrics to identify relative smoothness between successive peaks, time between peaks and overall smoothness indicators to quantify and compare between multiple cases. Our results demonstrate that EMMA can successfully identify the motion and wrinkles on each video frame and quantify the smoothness and identify the phases of each cardiac cycle. Moreover, EMMA can obtain the smoothness of each frame and the temporal evolution of membrane smoothness across all image frames for the video sequence. EMMA allows for a fast, accurate, quantitative assessment to be completed and reduces user error. This enables EMMA to be used effectively as an early warning system to rapidly identify device abnormalities.

L-PBF High-Throughput Data Pipeline Approach for Multi-modal Integration

Metal-based additive manufacturing requires active monitoring solutions for assessing part quality. Multiple sensors and data streams, however, generate large heterogeneous data sets that are impractical for manual assessment and characterization. In this work, an automated pipeline is developed that enables feature extraction from high-speed camera video and multi-modal data analysis. The framework removes the need for manual assessment through the utilization of deep learning techniques and training models in a weakly supervised paradigm. We demonstrate this pipeline’s capability over 700,000 high-speed camera frames. The pipeline successfully extracts melt pool and spatter geometries and links them to corresponding pyrometry, radiography, and processparameter information. 715 individual prints are examined to reveal melt pool areas that exceeds 0.07 mm2 and pyrometry signal over a threshold (375 pyrometry units) were more likely to have defects. These automated processes enable massive throughput of characterization techniques.

A Video Data Analysis of Pet Theft Incidents: An Examination of Offense Form, Situational Dynamics, & Offender Characteristics

ABSTRACT Pet theft is a type of property crime, but pet-owner relationships include emotional dimensions and life experiences that can compound their loss far beyond property value. Their bondedness with humans includes companionship as well as healthy lifestyle and mental wellness benefits, which are abruptly halted at junctures of pet loss. Though pets are stolen for various motives (e.g. ransom, reward, resale, or breeding), this crime form has received little research attention from academic criminology and criminal justice. Toward building a line of inquiry, this paper empirically explores the phenomena of “dognapping” as dogs are the most frequently stolen pet and the only pet for which there is a database large enough for meaningful exploration. Content analysis of online posts and web-based video data (Twitter/X1 posts, YouTube videos, and social media websites dedicated to missing and stolen pets) enabled a delineation of dog theft forms, offense dynamics, and offender characteristics. Findings regarding residential and commercial dog theft center discussion on situational crime prevention implications and additional research opportunities to further evidence and explore this relatively disregarded crime.

Optimizing Video Queries with Declarative Clues

Video Database Management Systems (VDBMS) leverage advancements in computer vision and deep learning for efficient video data analysis and retrieval. This paper introduces the concept of user-specified Clues, allowing users to incorporate domain-specific knowledge, referred to as Clues, into query optimization. Clues are expressed as Clue types, each associated with optimization rules, and applied to queries through Clue instances. The extensible ClueVQS system we present to incorporate these ideas, optimizes queries automatically, utilizing Clues to improve processing efficiency. We also introduce algorithms to optimize queries using Clues allowing for trade-offs between speed and query accuracy. Our proposals and system address challenges such as data-dependent Clue effectiveness, limiting search space, and accuracy-efficiency trade-offs. Detailed experimental results demonstrate query speedups of up to two orders of magnitude compared to other applicable approaches, and a reduction of the query optimizer time by up to 95% while respecting user-specified accuracy constraints, showcasing the effectiveness of the proposed framework.

Fluid viscosity prediction leveraging computer vision and robot interaction

Accurately determining fluid viscosity is crucial for various industrial and scientific applications. Traditional methods of viscosity measurement, though reliable, often require manual intervention and cannot easily adapt to real-time monitoring. With advancements in machine learning and computer vision, this work explores the feasibility of predicting fluid viscosity by analyzing fluid oscillations captured in video data. The pipeline employs a 3D convolutional autoencoder pretrained in a self-supervised manner to extract and learn features from semantic segmentation masks of oscillating fluids. Then, the latent representations of the input data, produced from the pretrained autoencoder, are processed with a distinct inference head to infer either the fluid category (classification) or the fluid viscosity (regression) in a time-resolved manner. When the latent representations generated by the pre-trained autoencoder are used for classification, the system achieves a 97.1% accuracy across a total of 4140 test datapoints. Similarly, for regression tasks, employing an additional fully-connected network as a regression head allows the pipeline to achieve a mean absolute error of 0.258 cP over 4416 test datapoints. This study represents an innovative contribution to both fluid characterization and the evolving landscape of Artificial Intelligence, demonstrating the potential of deep learning in achieving near real-time viscosity estimation and addressing practical challenges in fluid dynamics through the analysis of video data capturing oscillating fluid dynamics.

Beyond active learning: Using 3-Dimensional learning to create scientifically authentic, student-centered classrooms.

In recent years, much of the emphasis for transformation of introductory STEM courses has focused on "active learning", and while this approach has been shown to produce more equitable outcomes for students, the construct of "active learning" is somewhat ill-defined and is often used as a "catch-all" that can encompass a wide range of pedagogical techniques. Here we present an alternative approach for how to think about the transformation of STEM courses that focuses instead on what students should know and what they can do with that knowledge. This approach, known as three-dimensional learning (3DL), emerged from the National Academy's "A Framework for K-12 Science Education", which describes a vision for science education that centers the role of constructing productive causal accounts for phenomena. Over the past 10 years, we have collected data from introductory biology, chemistry, and physics courses to assess the impact of such a transformation on higher education courses. Here we report on an analysis of video data of class sessions that allows us to characterize these sessions as active, 3D, neither, or both 3D and active. We find that 3D classes are likely to also involve student engagement (i.e. be active), but the reverse is not necessarily true. That is, focusing on transformations involving 3DL also tends to increase student engagement, whereas focusing solely on student engagement might result in courses where students are engaged in activities that do not involve meaningful engagement with core ideas of the discipline.

Designing computational models as Emergent Systems Microworlds to support epistemically agentive learning of emergent biological phenomena

ABSTRACT To support students’ agency in the process of constructing knowledge in a science classroom, it is important to design learning environments that allow students to shape knowledge building practices. In this paper, we present an Emergent Systems Microworld (ESM)-based learning environment called GenEvo, which is designed to ‘restructurate’ learning of fundamental ideas in modern biology, such as gene regulation. We study how cognitive, social, and affective properties of agent-based restructurations supported student learning. We report findings from a qualitative analysis of video data of student participation and interviews in the fourth iteration of a design-based research project about an ESM-based curricular unit. We discuss how specific design features of the ESM supported students’ epistemically agentive learning. Students of the GenEvo course shaped practices to investigate and construct knowledge about emergent biological phenomena and learned about emergent phenomena related to gene regulation and evolution. This work demonstrates how the properties of a restructuration make ESM-based learning environments effective for students to collectively shape knowledge building practices and learn about emergent phenomena.

Fraternisation and repression during the 2020–2021 attempted revolution in Belarus

ABSTRACT Through adoption of Video Data Analysis (VDA), this study analyses the 2020–2021 revolutionary attempt in Belarus which featured the largest protests in the country's post-communist history. This study asks why state repression and violence broke out in some instances and interactions, but not in others? I analyse events that took place between activists and state forces through a sample of video data featuring 144 different cases including verbal interactions between protesters and authorities; fraternisation, and whether or not activists were able to strike a positive dialogue with police or security forces. These variables are statistically analysed with relation to whether authorities repressed protesters or if they successfully managed to put crowds under control. Results reveal that fraternisation and the establishment of a positive dialogue between protesters and state forces are negatively correlated with occurrences of state repression, which yields new insight into the micro-dynamics of repression and dissent.

Sources of Sound Exposure in Pediatric Critical Care.

Sound levels in the pediatric intensive care unit (PICU) are often above recommended levels, but few researchers have identified the sound sources contributing to high levels. To identify sources of PICU sound exposure. This was a secondary analysis of continuous bedside video and dosimeter data (n = 220.7 hours). A reliable coding scheme developed to identify sound sources in the adult ICU was modified for pediatrics. Proportions of sound sources were compared between times of high (≥45 dB) and low (<45 dB) sound, during day (7 AM to 6:59 PM) and night (7 PM to 6:59 AM) shifts, and during sound peaks (≥70 dB). Overall, family vocalizations (38% of observation time, n = 83.9 hours), clinician vocalizations (32%, n = 70.6 hours), and child nonverbal vocalizations (29.4%, n = 64.9 hours) were the main human sound sources. Media sounds (57.7%, n = 127.3 hours), general activity (40.7%, n = 89.8 hours), and medical equipment (31.3%, n = 69.1 hours) were the main environmental sound sources. Media sounds occurred in more than half of video hours. Child nonverbal (71.6%, n = 10.2 hours) and family vocalizations (63.2%, n = 9 hours) were highly prevalent during sound peaks. General activity (32.1%, n = 33.2 hours), clinician vocalizations (22.5%, n = 23.3 hours), and medical equipment sounds (20.6, n = 21.3 hours) were prevalent during night shifts. Clinicians should partner with families to limit nighttime PICU noise pollution. Large-scale studies using this reliable coding scheme are needed to understand the PICU sound environment.

World futures through RT’s eyes: multimodal dataset and interdisciplinary methodology

There is a need to develop new interdisciplinary approaches suitable for a more complete analysis of multimodal data. Such approaches need to go beyond case studies and leverage technology to allow for statistically valid analysis of the data. Our study addresses this need by engaging with the research question of how humans communicate about the future for persuasive and manipulative purposes, and how they do this multimodally. It introduces a new methodology for computer-assisted multimodal analysis of video data. The study also introduces the resulting dataset, featuring annotations for speech (textual and acoustic modalities) and gesticulation and corporal behaviour (visual modality). To analyse and annotate the data and develop the methodology, the study engages with 23 26-min episodes of the show ‘SophieCo Visionaries’, broadcast by RT (formerly ‘Russia Today’).