Search results for: video question answering

Authors: Haozheng Luo, Ruiyang Qin

Abstract:

People with visual impairments urgently need assistance, not only on the fundamental tasks such as guiding and retrieving objects but on the advanced like picturing the new environments. More than a guiding dog, they might want such devices that can provide linguistic interaction. Building on this idea, we aim to study the interaction between the robot agent and visually impaired people. In our research, we are going to develop a robot agent that will be able to analyze the test environment and answer the participants’ questions. We also will study the relevant issues regarding the interaction between human beings and the robot agents to figure out which and how the factors will affect the interaction.

Keywords: HRI, video question answering, visual question answering, natural language processing

Procedia PDF Downloads 189

Authors: Aya Mousa, Mahmoud Alsaheb

Abstract:

Question Answering System (QAS) aims to provide the most suitable answer to the user's question in any natural language. In the recent future, it will be a future version of web search. Much research has already been done on answering Arabic factoid questions and achieved good accuracy. In contrast, the progress in research on Arabic non-factoid question answering is still immature. In this survey, we summarize, discuss, and compare the existing Arab non-factoid question-answering systems to identify the limitations and the achievements that were accomplished. Furthermore, we investigate the challenges in developing non-factoid Arabic QAS and the possible future improvements. The survey is written to help the researchers to understand the field of Arabic non-factoid QAS and to motivate them to utilize different approaches to develop and enhance the Non-factoid Arabic QAS

Keywords: Arabic question answering system, non-factoid question answering, Arabic NLP, question answering

Procedia PDF Downloads 55

Authors: Touila Ahmed, Elie Louis, Hamza Gharbi

Abstract:

State of the art question answering models have recently shown impressive performance especially in a zero-shot setting. This approach is particularly useful when confronted with a highly diverse domain such as the legal field, in which it is increasingly difficult to have a dataset covering every notion and concept. In this work, we propose a flexible generative question answering approach to contract analysis as well as a weakly supervised procedure to leverage unannotated data and boost our models’ performance in general, and their zero-shot performance in particular.

Keywords: question answering, contract analysis, zero-shot, natural language processing, generative models, self-supervision

Procedia PDF Downloads 142

Authors: Wided Bakari, Patrice Bellot, Omar Trigui, Mahmoud Neji

Abstract:

This article constitutes an opening to think of the modeling and analysis of Arabic texts in the context of a question-answer system. It is a question of exceeding the traditional approaches focused on morphosyntactic approaches. Furthermore, we present a new approach that analyze a text in order to extract correct answers then transform it to logical predicates. In addition, we would like to represent different levels of information within a text to answer a question and choose an answer among several proposed. To do so, we transform both the question and the text into logical forms. Then, we try to recognize all entailment between them. The results of recognizing the entailment are a set of text sentences that can implicate the user’s question. Our work is now concentrated on an implementation step in order to develop a system of question-answering in Arabic using techniques to recognize textual implications. In this context, the extraction of text features (keywords, named entities, and relationships that link them) is actually considered the first step in our process of text modeling. The second one is the use of techniques of textual implication that relies on the notion of inference and logic representation to extract candidate answers. The last step is the extraction and selection of the desired answer.

Keywords: NLP, Arabic language, question-answering, recognition text entailment, logic forms

Procedia PDF Downloads 309

Authors: Qisheng Hu, Jianglei Han, Yue Yang, My Hoa Ha

Abstract:

When a company is considering purchasing software for commercial use, contract risk assessment is critical to identify risks to mitigate the potential adverse business impact, e.g., security, financial and regulatory risks. Contract risk assessment requires reviewers with specialized knowledge and time to evaluate the legal documents manually. Specifically, validating contracts for a software vendor requires the following steps: manual screening, interpreting legal documents, and extracting risk-prone segments. To automate the process, we proposed a framework to assist legal contract document risk identification, leveraging pre-trained deep learning models and natural language processing techniques. Given a set of pre-defined risk evaluation problems, our framework utilizes the pre-trained transformer-based models for question-answering to identify risk-prone sections in a contract. Furthermore, the question-answering model encodes the concatenated question-contract text and predicts the start and end position for clause extraction. Due to the limited labelled dataset for training, we leveraged transfer learning by fine-tuning the models with the CUAD dataset to enhance the model. On a dataset comprising 287 contract documents and 2000 labelled samples, our best model achieved an F1 score of 0.687.

Keywords: contract risk assessment, NLP, transfer learning, question answering

Procedia PDF Downloads 93

Authors: Prathiksha Kamath, Pratibha Jamkhandi, Prateek Ghanti, Priyanshu Gupta, M. Lakshmi Neelima

Abstract:

Combining Image Captioning and Visual Question Answering (VQA) tasks have emerged as a new and exciting research area. The image captioning task involves generating a textual description that summarizes the content of the image. VQA aims to answer a natural language question about the image. Both these tasks include computer vision and natural language processing (NLP) and require a deep understanding of the content of the image and semantic relationship within the image and the ability to generate a response in natural language. There has been remarkable growth in both these tasks with rapid advancement in deep learning. In this paper, we present a comprehensive review of recent progress in combining image captioning and visual question-answering (VQA) tasks. We first discuss both image captioning and VQA tasks individually and then the various ways in which both these tasks can be integrated. We also analyze the challenges associated with these tasks and ways to overcome them. We finally discuss the various datasets and evaluation metrics used in these tasks. This paper concludes with the need for generating captions based on the context and captions that are able to answer the most likely asked questions about the image so as to aid the VQA task. Overall, this review highlights the significant progress made in combining image captioning and VQA, as well as the ongoing challenges and opportunities for further research in this exciting and rapidly evolving field, which has the potential to improve the performance of real-world applications such as autonomous vehicles, robotics, and image search.

Keywords: image captioning, visual question answering, deep learning, natural language processing

Procedia PDF Downloads 46

Authors: Igor A. Podgorny

Abstract:

TurboTax AnswerXchange is a social Q/A system supporting users working on federal and state tax returns. Content quality and popularity in the AnswerXchange can be predicted with propensity models using attributes of the question and answer. Using uplift modeling, we identify features of questions and answers that can be modified during the question-asking and question-answering experience in order to optimize the AnswerXchange content quality. We demonstrate that adding details to the questions always results in increased question popularity that can be used to promote good quality content. Responding to close-ended questions assertively improve content quality in the AnswerXchange in 90% of cases. Answering knowledge questions with web links increases the likelihood of receiving a negative vote from 60% of the askers. Our findings provide a rationale for employing the uplift modeling approach for AnswerXchange operations.

Keywords: customer relationship management, human-machine interaction, text mining, uplift modeling

Procedia PDF Downloads 222

Authors: Anes Enakoa, Yawei Liang

Abstract:

Question Answering (QA) system is one of the most important and demanding tasks in the field of Natural Language Processing (NLP). In QA systems, the answer generation task generates a list of candidate answers to the user's question, in which only one answer is correct. Answer selection is one of the main components of the QA, which is concerned with selecting the best answer choice from the candidate answers suggested by the system. However, the selection process can be very challenging especially in Arabic due to its particularities. To address this challenge, an approach is proposed to answer questions with multiple answer choices for Arabic QA systems based on Textual Entailment (TE) recognition. The developed approach employs a Support Vector Machine that considers lexical, semantic and syntactic features in order to recognize the entailment between the generated hypotheses (H) and the text (T). A set of experiments has been conducted for performance evaluation and the overall performance of the proposed method reached an accuracy of 67.5% with C@1 score of 80.46%. The obtained results are promising and demonstrate that the proposed method is effective for TE recognition task.

Keywords: information retrieval, machine learning, natural language processing, question answering, textual entailment

Procedia PDF Downloads 115

Authors: Aviv Yaniv, Ron Ben Arosh, Nadav Gasner, Michael Konviser, Arbel Yaniv

Abstract:

This paper introduces two methods for the evaluation of Question Decomposition Meaning Representation (QDMR) as predicted by sequence-to-sequence model and COPYNET parser for natural language questions processing, motivated by the fact that previous evaluation metrics used for this task do not take into account some characteristics of the representation, such as partial ordering structure. To this end, several heuristics to extract such partial dependencies are formulated, followed by the hereby proposed evaluation methods denoted as Proportional Graph Matcher (PGM) and Conversion to Normal String Representation (Nor-Str), designed to better capture the accuracy level of QDMR predictions. Experiments are conducted to demonstrate the efficacy of the proposed evaluation methods and show the added value suggested by one of them- the Nor-Str, for better distinguishing between high and low-quality QDMR when predicted by models such as COPYNET. This work represents an important step forward in the development of better evaluation methods for QDMR predictions, which will be critical for improving the accuracy and reliability of natural language question-answering systems.

Keywords: NLP, question answering, question decomposition meaning representation, QDMR evaluation metrics

Procedia PDF Downloads 34

Authors: Long Chen

Abstract:

With the proliferation of E-Learning, collaborative learning becomes more and more popular in various teaching and learning occasions. Studies over the years have proved that actively participating in classroom discussion can enhance student's learning experience, consolidating their knowledge and understanding of the class content. Collaborative learning can also allow students to share their resources and knowledge by exchanging, absorbing, and observing one another's opinions and ideas. Community Question Answering (CQA) services are particularly suitable paradigms for collaborative learning, since it is essentially an online collaborative learning platform where one can get information from multiple sources for he/her to choose from. However, current CQA services have only achieved limited success in collaborative learning due to the uncertainty of answers' quality. In this paper, we predict the quality of answers in a CQA service, i.e. Yahoo! Answers, for the use of online education and distance learning, which would enable a student to find relevant answers and potential answerers more effectively and efficiently, and thus greatly increase students' user experience in CQA services. Our experiment reveals that the quality of answers is influenced by a series of factors such as asking time, relations between users, and his/her experience in the past. We also show that by modelling user's profile with our proposed personalized features, student's satisfaction towards the provided answers could be accurately estimated.

Keywords: Community Question Answering, Collaborative Learning, Log File, Co-Training

Procedia PDF Downloads 413

Authors: Won-Ho Kim

Abstract:

In this paper, a design of H.263 based wireless video transceiver is presented for wireless camera system. It uses standard WIFI transceiver and the covering area is up to 100m. Furthermore the standard H.263 video encoding technique is used for video compression since wireless video transmitter is unable to transmit high capacity raw data in real time and the implemented system is capable of streaming at speed of less than 1Mbps using NTSC 720x480 video.

Keywords: wireless video transceiver, video surveillance camera, H.263 video encoding digital signal processing

Procedia PDF Downloads 335

Authors: Amarjit Singh

Abstract:

In this paper, a method for extraction of text subtitles in large video is proposed. The video data needs to be annotated for many multimedia applications. Text is incorporated in digital video for the motive of providing useful information about that video. So need arises to detect text present in video to understanding and video indexing. This is achieved in two steps. First step is text localization and the second step is text verification. The method of text detection can be extended to text recognition which finds applications in automatic video indexing; video annotation and content based video retrieval. The method has been tested on various types of videos.

Keywords: video, subtitles, extraction, annotation, frames

Procedia PDF Downloads 566

Authors: Zaynab El Khattabi, Youness Tabii, Abdelhamid Benkaddour

Abstract:

Nowadays, huge amount of multimedia repositories make the browsing, retrieval and delivery of video contents very slow and even difficult tasks. Video summarization has been proposed to improve faster browsing of large video collections and more efficient content indexing and access. In this paper, we focus on approaches to video summarization. The video summaries can be generated in many different forms. However, two fundamentals ways to generate summaries are static and dynamic. We present different techniques for each mode in the literature and describe some features used for generating video summaries. We conclude with perspective for further research.

Keywords: video summarization, static summarization, video skimming, semantic features

Procedia PDF Downloads 369

Authors: Shraddha Chaudhary, Raksha Agarwal, Niladri Chatterjee

Abstract:

This work demonstrates a web crawler-based generalized end-to-end open domain Question Answering (QA) system. An efficient QA system requires a significant amount of domain knowledge to answer any question with the aim to find an exact and correct answer in the form of a number, a noun, a short phrase, or a brief piece of text for the user's questions. Analysis of the question, searching the relevant document, and choosing an answer are three important steps in a QA system. This work uses a web scraper (Beautiful Soup) to extract K-documents from the web. The value of K can be calibrated on the basis of a trade-off between time and accuracy. This is followed by a passage ranking process using the MS-Marco dataset trained on 500K queries to extract the most relevant text passage, to shorten the lengthy documents. Further, a QA system is used to extract the answers from the shortened documents based on the query and return the top 3 answers. For evaluation of such systems, accuracy is judged by the exact match between predicted answers and gold answers. But automatic evaluation methods fail due to the linguistic ambiguities inherent in the questions. Moreover, reference answers are often not exhaustive or are out of date. Hence correct answers predicted by the system are often judged incorrect according to the automated metrics. One such scenario arises from the original Google Natural Question (GNQ) dataset which was collected and made available in the year 2016. Use of any such dataset proves to be inefficient with respect to any questions that have time-varying answers. For illustration, if the query is where will be the next Olympics? Gold Answer for the above query as given in the GNQ dataset is “Tokyo”. Since the dataset was collected in the year 2016, and the next Olympics after 2016 were in 2020 that was in Tokyo which is absolutely correct. But if the same question is asked in 2022 then the answer is “Paris, 2024”. Consequently, any evaluation based on the GNQ dataset will be incorrect. Such erroneous predictions are usually given to human evaluators for further validation which is quite expensive and time-consuming. To address this erroneous evaluation, the present work proposes an automated approach for evaluating time-dependent question-answer pairs. In particular, it proposes a metric using the current timestamp along with top-n predicted answers from a given QA system. To test the proposed approach GNQ dataset has been used and the system achieved an accuracy of 78% for a test dataset comprising 100 QA pairs. This test data was automatically extracted using an analysis-based approach from 10K QA pairs of the GNQ dataset. The results obtained are encouraging. The proposed technique appears to have the possibility of developing into a useful scheme for gathering precise, reliable, and specific information in a real-time and efficient manner. Our subsequent experiments will be guided towards establishing the efficacy of the above system for a larger set of time-dependent QA pairs.

Keywords: web-based information retrieval, open domain question answering system, time-varying QA, QA evaluation

Procedia PDF Downloads 71

Authors: Van Thinh Bui, Anh Tuan Tran, Quoc Viet Ngo, The Bao Pham

Abstract:

Video retrieval is a searching problem on videos or clips based on content in which they are relatively close to an input image or video. The application of this retrieval consists of selecting video in a folder or recognizing a human in security camera. However, some recent approaches have been in challenging problem due to the diversity of video types, frame transitions and camera positions. Besides, that an appropriate measures is selected for the problem is a question. In order to overcome all obstacles, we propose a content-based video retrieval system in some main steps resulting in a good performance. From a main video, we process extracting keyframes and principal objects using Segmentation of Aggregating Superpixels (SAS) algorithm. After that, Speeded Up Robust Features (SURF) are selected from those principal objects. Then, the model “Bag-of-words” in accompanied by SVM classification are applied to obtain the retrieval result. Our system is performed on over 300 videos in diversity from music, history, movie, sports, and natural scene to TV program show. The performance is evaluated in promising comparison to the other approaches.

Keywords: video retrieval, principal objects, keyframe, segmentation of aggregating superpixels, speeded up robust features, bag-of-words, SVM

Procedia PDF Downloads 272

Authors: B. J. Sandesh, Saurabha Jirgi, S. Vidya, Prakash Eljer, Gowri Srinivasa

Abstract:

In this paper, we propose a framework to help users to search and retrieve the portions in the lecture video of their interest. This is achieved by temporally segmenting and indexing the lecture video using the topic keywords. We use transcribed text from the video and documents relevant to the video topic extracted from the web for this purpose. The keywords for indexing are found by applying the non-negative matrix factorization (NMF) topic modeling techniques on the web documents. Our proposed technique first creates indices on the transcribed documents using the topic keywords, and these are mapped to the video to find the start and end time of the portions of the video for a particular topic. This time information is stored in the index table along with the topic keyword which is used to retrieve the specific portions of the video for the query provided by the users.

Keywords: video indexing and retrieval, lecture videos, content based video search, multimodal indexing

Procedia PDF Downloads 204

Authors: U. S. N. Raju, Kothuri Sai Kiran, Meena G. Kamal, Vinay Nikhil Pabba, Suresh Kanaparthi

Abstract:

There is huge amount of lecture video data available for public use, and many more lecture videos are being created and uploaded every day. Searching for videos on required topics from this huge database is a challenging task. Therefore, an efficient method for video retrieval is needed. An approach for automated video indexing and video search in large lecture video archives is presented. As the amount of video lecture data is huge, it is very inefficient to do the processing in a centralized computation framework. Hence, Hadoop Framework for distributed computing for Big Video Data is used. First, step in the process is automatic video segmentation and key-frame detection to offer a visual guideline for the video content navigation. In the next step, we extract textual metadata by applying video Optical Character Recognition (OCR) technology on key-frames. The OCR and detected slide text line types are adopted for keyword extraction, by which both video- and segment-level keywords are extracted for content-based video browsing and search. The performance of the indexing process can be improved for a large database by using distributed computing on Hadoop framework.

Keywords: video lectures, big video data, video retrieval, hadoop

Procedia PDF Downloads 491

Authors: Norah Alshayhan

Abstract:

Dealing with public problems is one of the struggles that may face the leaders in the public sector. Collaborative leadership is one of the most important approaches in dealing with difficult situations that affect both public, private, and nonprofit organizations. Current literature does not show exactly the extent of utilizing collaborative leadership during the post-COVID-19 world in Saudi Arabia. This study is worth exploring in order to examine collaborative leadership in similar environments. This research will utilize both integrative public leadership and transformational leadership theories to guide the researcher in answering the research question. The researcher utilizes content analysis and reviews government documents, plans, and reports to gain more information about collaborative leadership in Saudi Arabia. The researcher analyzes the data in themes and sub-themes to categorize the data in answering the research question. Leader’s behavior and performance in the public sector will be the focus of this study. Findings from this research will benefit leaders in both public, private, and nonprofit sectors in their leadership during a post-disaster time. Findings from this study support collaborative leadership practices and performance in leading future post-crises/disasters.

Keywords: collaborative leadership, post-COVID-19, Saudi Arabia, performance, skills

Procedia PDF Downloads 37

Authors: Shamsundar Kulkarni

Abstract:

Video capturing by non-professionals will lead to unanticipated effects. Such as image distortion, image blurring etc. Hence, many researchers study such drawbacks to enhance the quality of videos. In this paper, an algorithm is proposed to stabilize jittery videos .A stable output video will be attained without the effect of jitter which is caused due to shaking of handheld camera during video recording. Firstly, salient points from each frame from the input video are identified and processed followed by optimizing and stabilize the video. Optimization includes the quality of the video stabilization. This method has shown good result in terms of stabilization and it discarded distortion from the output videos recorded in different circumstances.

Keywords: video stabilization, point feature matching, salient points, image quality measurement

Procedia PDF Downloads 275

Authors: Qin Luofeng, Shen Siqi

Abstract:

For the 58 years since the first video game came into being, the video game industry is getting through an explosive evolution from then on. Video games exert great influence on society and become a reflection of public life to some extent. Video game virtual spaces are where activities are taking place like real spaces. And that’s the reason why some architects pay attention to video games. However, compared to the researches on the appearance of games, we observe a lack of theoretical comprehensive on the construction of video game virtual spaces. The research method of this paper is to collect literature and conduct theoretical research about the virtual space in video games firstly. And then analogizing the opinions on the space phenomena from the theory of literature and films. Finally, this paper proposes a three-layer framework for the construction of video game virtual spaces: “algorithmic space-narrative space players space”, which correspond to the exterior, expressive, affective parts of the game space. Also, we illustrate each sub-space according to numerous instances of published video games. Hoping this writing could promote the interactive development of video games and architecture.

Keywords: video game, virtual space, narrativity, social space, emotional connection

Procedia PDF Downloads 223

Authors: Janya Sainui

Abstract:

As a rapid growth of digital videos and data communications, video summarization that provides a shorter version of the video for fast video browsing and retrieval is necessary. Key frame extraction is one of the mechanisms to generate video summary. In general, the extracted key frames should both represent the entire video content and contain minimum redundancy. However, most of the existing approaches heuristically select key frames; hence, the selected key frames may not be the most different frames and/or not cover the entire content of a video. In this paper, we propose a method of video summarization which provides the reasonable objective functions for selecting key frames. In particular, we apply a statistical dependency measure called quadratic mutual informaion as our objective functions for maximizing the coverage of the entire video content as well as minimizing the redundancy among selected key frames. The proposed key frame extraction algorithm finds key frames as an optimization problem. Through experiments, we demonstrate the success of the proposed video summarization approach that produces video summary with better coverage of the entire video content while less redundancy among key frames comparing to the state-of-the-art approaches.

Keywords: video summarization, key frame extraction, dependency measure, quadratic mutual information

Procedia PDF Downloads 244

Authors: Soo Hyung Kim

Abstract:

Two different learning approaches were suggested: focusing on factual knowledge or focusing on the embedded meaning in the statements. Each way of learning has positive effects on different question categories, where factual knowledge helps more with simple fact questions, and searching for meaning in given information helps learn causal relationship and the embedded meaning. To test this belief, two groups of learners (12 male and 39 female adults aged 18-37) watched a ten-minute long Youtube video about various factual events of American history, their meaning, and the causal relations of the events. The fact group was asked to focus on factual knowledge in the video, and the meaning group was asked to focus on the embedded meaning in the video. After watching the video, both groups took multiple-choice questions, which consisted of 10 questions asking the factual knowledge addressed in the video and 10 questions asking embedded meaning in the video, such as the causal relationship between historical events and the significance of the event. From ANCOVA analysis, it was found that the factual knowledge showed higher performance on the factual questions than the meaning group, although there was no group difference on the questions about the meaning between the two groups. The finding suggests that teacher instruction plays an important role in learners constructing a different type of knowledge in online learning.

Keywords: factual knowledge, instruction, meaning-based knowledge, online learning

Procedia PDF Downloads 109

Authors: S. Sarumathi, C. Deepadharani, Garimella Archana, S. Dakshayani, D. Logeshwaran, D. Jayakumar, Vijayarangan Natarajan

Abstract:

The need of the hour for the customers who use a dial-up or a low broadband connection for their internet services is to access HD video data. This can be achieved by developing a new video format using H. 265. This is the latest video codec standard developed by ISO/IEC Moving Picture Experts Group (MPEG) and ITU-T Video Coding Experts Group (VCEG) on April 2013. This new standard for video compression has the potential to deliver higher performance than the earlier standards such as H. 264/AVC. In comparison with H. 264, HEVC offers a clearer, higher quality image at half the original bitrate. At this lower bitrate, it is possible to transmit high definition videos using low bandwidth. It doubles the data compression ratio supporting 8K Ultra HD and resolutions up to 8192×4320. In the proposed model, we design a new video format which supports this H. 265 standard. The major areas of applications in the coming future would lead to enhancements in the performance level of digital television like Tata Sky and Sun Direct, BluRay Discs, Mobile Video, Video Conferencing and Internet and Live Video streaming.

Keywords: access HD video, H. 265 video standard, high performance, high quality image, low bandwidth, new video format, video streaming applications

Procedia PDF Downloads 328

Authors: Assma Azeroual, Karim Afdel, Mohamed El Hajji, Hassan Douzi

Abstract:

Key frame extraction methods select the most representative frames of a video, which can be used in different areas of video processing such as video retrieval, video summary, and video indexing. In this paper we present a novel approach for extracting key frames from video sequences. The frame is characterized uniquely by his contours which are represented by the dominant blocks. These dominant blocks are located on the contours and its near textures. When the video frames have a noticeable changement, its dominant blocks changed, then we can extracte a key frame. The dominant blocks of every frame is computed, and then feature vectors are extracted from the dominant blocks image of each frame and arranged in a feature matrix. Singular Value Decomposition is used to calculate sliding windows ranks of those matrices. Finally the computed ranks are traced and then we are able to extract key frames of a video. Experimental results show that the proposed approach is robust against a large range of digital effects used during shot transition.

Keywords: FSDWT, key frame extraction, shot detection, singular value decomposition

Procedia PDF Downloads 353

Authors: Yagya Raj Pandeya, Joonwhoan Lee

Abstract:

The dynamic behavior of music and video makes it difficult to evaluate musical instrument playing in a video by computer system. Any television or film video clip with music information are rich sources for analyzing musical instruments using modern machine learning technologies. In this research, we integrate the audio and video information sources using convolutional neural network (CNN) and pass network learned features through recurrent neural network (RNN) to preserve the dynamic behaviors of audio and video. We use different pre-trained CNN for music and video feature extraction and then fine tune each model. The music network use 2D convolutional network and video network use 3D convolution (C3D). Finally, we concatenate each music and video feature by preserving the time varying features. The long short term memory (LSTM) network is used for long-term dynamic feature characterization and then use late fusion with generalized mean. The proposed network performs better performance to recognize the musical instrument using audio-video multimodal neural network.

Keywords: multimodal, 3D convolution, music-video feature extraction, generalized mean

Procedia PDF Downloads 185

Authors: C. von Essen

Abstract:

This study outlines and evaluates a constructivist design approach to the creation of educational video on postgraduate business degree programmes. Many online courses are tapping into the educational affordances of video, as this form of online learning has the potential to create rich, multimodal experiences. And yet, in many learning contexts video is still being used to transmit instruction to passive learners, rather than promote learner engagement and knowledge creation. Constructivism posits the notion that learning is shaped as students make connections between their experiences and ideas. This paper pivots on the following research question: how can we design educational video in ways which promote constructivist learning and stimulate analytic viewing? By exploring and categorizing over two thousand educational videos created since 2014 for over thirty postgraduate courses in business, economics, mathematics and statistics, this paper presents and critically reflects on a taxonomy of video styles and features. It links the pedagogical intent of video – be it concept explanation, skill demonstration, feedback, real-world application of ideas, community creation, or the cultivation of course narrative – to specific presentational characteristics such as visual effects including diagrammatic and real-life graphics and aminations, commentary and sound options, chronological sequencing, interactive elements, and presenter set-up. The findings of this study inform a framework which captures the pedagogical, technological and production considerations instructional designers and educational media specialists should be conscious of when planning and preparing the video. More broadly, the paper demonstrates how learning theory and technology can coalesce to produce informed and pedagogical grounded instructional design choices. This paper reveals how crafting video in a more conscious and critical manner can produce powerful, new educational design.

Keywords: educational video, constructivism, instructional design, business education

Procedia PDF Downloads 203

Authors: Nada Jasim Habeeb, Rana Saad Mohammed, Muntaha Khudair Abbass

Abstract:

For more efficient and fast video summarization, this paper presents a surveillance video summarization method. The presented method works to improve video summarization technique. This method depends on temporal differencing to extract most important data from large video stream. This method uses histogram differencing and Sum Conditional Variance which is robust against to illumination variations in order to extract motion objects. The experimental results showed that the presented method gives better output compared with temporal differencing based summarization techniques.

Keywords: temporal differencing, video summarization, histogram differencing, sum conditional variance

Procedia PDF Downloads 319

Authors: Kurt Maxwell Kusterer, Sylvain Mignot, Annick Vignes

Abstract:

To our best knowledge YouTube is the largest video hosting platform in the world. A high number of creators, viewers, subscribers and commentators act in this specific eco-system which generates huge sums of money. Views, subscribers, and comments help to increase the popularity of content creators. The most popular creators are sponsored by brands and participate in marketing campaigns. For a few of them, this becomes a financially rewarding profession. This is made possible through the YouTube Partner Program, which shares revenue among creators based on their popularity. We believe that the role of comments in increasing the popularity is to be emphasized. In what follows, YouTube is considered as a bilateral network between the videos and the commentators. Analyzing a detailed data set focused on French YouTubers, we consider each comment as a link between a commentator and a video. Our research question asks what are the predominant features of a video which give it the highest probability to be commented on. Following on from this question, how can we use these features to predict the action of the agent in commenting one video instead of another, considering the characteristics of the commentators, videos, topics, channels, and recommendations. We expect to see that the videos of more popular channels generate higher viewer engagement and thus are more frequently commented. The interest lies in discovering features which have not classically been considered as markers for popularity on the platform. A quick view of our data set shows that 96% of the commentators comment only once on a certain video. Thus, we study a non-weighted bipartite network between commentators and videos built on the sub-sample of 96% of unique comments. A link exists between two nodes when a commentator makes a comment on a video. We run an Exponential Random Graph Model (ERGM) approach to evaluate which characteristics influence the probability of commenting a video. The creation of a link will be explained in terms of common video features, such as duration, quality, number of likes, number of views, etc. Our data is relevant for the period of 2020-2021 and focuses on the French YouTube environment. From this set of 391 588 videos, we extract the channels which can be monetized according to YouTube regulations (channels with at least 1000 subscribers and more than 4000 hours of viewing time during the last twelve months).In the end, we have a data set of 128 462 videos which consist of 4093 channels. Based on these videos, we have a data set of 1 032 771 unique commentators, with a mean of 2 comments per a commentator, a minimum of 1 comment each, and a maximum of 584 comments.

Keywords: YouTube, social networks, economics, consumer behaviour

Procedia PDF Downloads 44

Authors: Maryam Azimi, Amin Banitalebi-Dehkordi, Yuanyuan Dong, Mahsa T. Pourazad, Panos Nasiopoulos

Abstract:

While there exists a wide variety of Low Dynamic Range (LDR) quality metrics, only a limited number of metrics are designed specifically for the High Dynamic Range (HDR) content. With the introduction of HDR video compression standardization effort by international standardization bodies, the need for an efficient video quality metric for HDR applications has become more pronounced. The objective of this study is to compare the performance of the existing full-reference LDR and HDR video quality metrics on HDR content and identify the most effective one for HDR applications. To this end, a new HDR video data set is created, which consists of representative indoor and outdoor video sequences with different brightness, motion levels and different representing types of distortions. The quality of each distorted video in this data set is evaluated both subjectively and objectively. The correlation between the subjective and objective results confirm that VIF quality metric outperforms all to their tested metrics in the presence of the tested types of distortions.

Keywords: HDR, dynamic range, LDR, subjective evaluation, video compression, HEVC, video quality metrics

Procedia PDF Downloads 485

Authors: Sikiru Ademola Adewale, Joe Thomas, Bolanle Hafiz Matti, Tosin Ige

Abstract:

This project demonstrates the implementation and use of an encoder-decoder model to perform a many-to-many mapping of video data to text captions. The many-to-many mapping occurs via an input temporal sequence of video frames to an output sequence of words to form a caption sentence. Data preprocessing, model construction, and model training are discussed. Caption correctness is evaluated using 2-gram BLEU scores across the different splits of the dataset. Specific examples of output captions were shown to demonstrate model generality over the video temporal dimension. Predicted captions were shown to generalize over video action, even in instances where the video scene changed dramatically. Model architecture changes are discussed to improve sentence grammar and correctness.

Keywords: decoder, encoder, many-to-many mapping, video captioning, 2-gram BLEU

Procedia PDF Downloads 63