<title>Automated video summarization using speech transcripts</title>

Abstract

Compact representations of video data can enable efficient video browsing. Such representations provide the user with information about the content of the particular sequence being examined while preserving the essential message. We propose a method to automatically generate video summaries for long videos. Our video summarization approach involves mainly two tasks: first, segmenting the video into small, coherent segments and second, ranking the resulting segments. Our proposed algorithm scores segments based on word frequency analysis of speech transcripts. Then a summary is generated by selecting the segments with the highest score to duration ratios and these are concatenating them. We have designed and performed a user study to evaluate the quality of summaries generated. Comparisons are made using our proposed algorithm and a random segment selection scheme based on statistical analysis of the user study results. Finally we discuss various issues that arise in summary evaluation with user studies.

Evaluating audio skimming and frame rate acceleration for summarizing BBC rushes

“…Wildemuth et al, note that fast-forward surrogates should ideally be controllable by end users who can adjust the speed based on content characteristics and personal preferences [10]. Interactive video summaries have received emphasis by others as well, e.g., in [8] the authors note the following, while acknowledging the difficulties of setting up assessment frameworks for such interactive, dynamic summaries:…”

“…RE and EA were included as metrics to help with assessing utility and end-user satisfaction, but while 25x was acknowledged as redundant, its ease of use measure (EA) was actually better than that for cluster. Such conflicts in assessing video summaries are discussed further in [8]: optimizing for one parameter like ease of use often comes at the expense of another like redundancy. The NIST overview report notes another such conflict with worse RE often leading to better IN: "redundancy does seem to make it more likely the ground truth items will be included and found…perhaps because it makes the assessor's job easier" [6].…”

“…The TRECVID pilot organizers define a summary as presenting "a condensed version of some information, such that various judgments about the full information can be made using only the summary and taking less time and effort than would be required using the full information source" [6]. This type of video summary is meant to serve both an indicative and informative function as defined in [8], giving the video summary all of the important information contained in the full information source. In a world of information overload, summaries have widespread application as compact surrogates returned by searches as previews, or used to give someone an efficient overview of a vast or unfamiliar video collection [6].…”

“…Video summaries can help direct viewers to relevant content, saving effort, network resources, and increasing end user satisfaction. Video summaries can take many forms, for example as text labels, single thumbnails, storyboards of thumbnails, or dynamic slide shows [8,9]. This paper concentrates on playable video summaries, experimenting with summaries that have durations of one-twenty-fifth (4%) or smaller compared to the target video.…”

Comparativo entre fontes de dados para anotação automática de videoaulas

“…Estes tipos de vídeos não podem ser facilmente segmentados de acordo com um evento específico, cada parte do vídeo se mostra igualmente importante para o usuário [Taskiran et al 2006]. Neste contexto, abordagens que fazem uso de reconhecimento de fala, como a apresentada em [Raimond and Lowis 2012] podem gerar informaçãoútil para a anotação.…”