The assignment of importance scores to particular frames or (short) segments in a video is crucial for summarization, but also difficult task. Previous work utilizes only one source visual features. In this paper, we suggest novel model architecture that combines three feature sets content and motion predict scores. proposed an attention mechanism before fusing features representing the (static) content, i.e., derived from image classification model. Comprehensive experimental evaluations...

Video summarization aims at generating a compact yet representative visual summary that conveys the essence of original video. The advantage unsupervised approaches is they do not require human annotations to learn capability and generalize wider range domains. Previous work relies on same type deep features, typically based model pre-trained ImageNet data. Therefore, we propose incorporate multiple feature sources with chunk stride fusion provide more information about content. For...

Videos are a commonly-used type of content in learning during Web search. Many e-learning platforms provide quality content, but sometimes educational videos long and cover many topics. Humans good extracting important sections from videos, it remains significant challenge for computers. In this paper, we address the problem assigning importance scores to video segments, that is how much information they contain with respect overall topic an video. We present annotation tool new dataset...

Cultural research is dedicated to understanding the processes of knowledge dissemination and social technological practices in book industry. Research on children books 19th century can be supported by computer systems. Specifically, advances digital image processing seem offer great opportunities for analyzing quantifying visual components books. The production technology illustrations was characterized a shift from wood or copper engraving lithography. We report classification experiments...

Due to the swift growth of patent applications each year, information and multimedia retrieval approaches that facilitate exploration are utmost importance. Different types visualizations (e.g., graphs, technical drawings) perspectives side view, perspective) used visualize details innovations in patents. The classification these images enables a more efficient search allows for further analysis. So far, datasets image type miss some important visualization Furthermore, related work does not...

Analysis of smart devices’ sensor data for the classification human activities has become increasingly targeted by industry and motion research. With popularization smartwatches, this becomes available to everyone. The user’s from accelerometers gyroscopes is conventionally analyzed as a multivariate time series obtain reliable information about activity at specific moment. Due particular sampling rate instabilities each device, previous approaches mainly work with feature extraction methods...

Video summarization aims at generating a compact yet representative visual summary that conveys the essence of original video. The advantage unsupervised approaches is they do not require human annotations to learn capability and generalize wider range domains. Previous work relies on same type deep features, typically based model pre-trained ImageNet data. Therefore, we propose incorporation multiple feature sources with chunk stride fusion provide more information about content. For...

The assignment of importance scores to particular frames or (short) segments in a video is crucial for summarization, but also difficult task. Previous work utilizes only one source visual features. In this paper, we suggest novel model architecture that combines three feature sets content and motion predict scores. proposed an attention mechanism before fusing features representing the (static) content, i.e., derived from image classification model. Comprehensive experimental evaluations...