Proceedings of the 1st ACM International Workshop on Human Centered Event Understanding from Multimedia

Abstract

It is our great pleasure to welcome you to the 1st ACM International Workshop on Humancentered Event Understanding from Multimedia1 - HuEvent'14. This year's workshop continues its tradition of being the premier ACM workshop for presentation of research results and experience reports in the field of events in multimedia. The mission of the workshop is to present and discuss the different aspects and notions of events and objects. This includes methods for detecting activities and low-level events and objects from media content and other sensory data. It also targets solutions and approaches for detecting and modeling the relationships between events and objects. The workshop is organized into a keynote, two sets of paper presentations, and a final discussion. The keynote will be held by Ramesh Jain, University of California at Irvine, CA, USA, with the title the Objective Self. Subsequently, a set of papers will be presented that deal with the detection of events in video. In this case, an event is most often defined as a complex interaction pattern that involves people and nobjects and is depicted in the video. In Entity centric Feature Pooling for Complex Event Detection, the authors tackle the problem of detecting such events in YouTube-like videos. Their approach is based on understanding the spatial arrangement of people and objects that participate to an event. In Skeleton-augmented Human Action Understanding by Learning with Progressively Refined Data, the authors focus on the human actors of the event, specifically on inferring skeletons that describe the human actions depicted in the video. In Using Minute-by-Minute Match Report for Semantic Event Annotation in Soccer Video, the event-based analysis of video belonging to a very specific domain (soccer matches) is addressed. The soccer video is annotated using the match report; however, without assuming that perfect time synchronization exists between the video content and textual description of the match. Considering again a very specific but radically different domain, the authors of Event Understanding in Endoscopic Surgery Videos examine the segmentation of surgery video in sub-events. The second set of papers takes a somewhat different view on events: Instead of looking at events as short-term human actions or human-object interactions that are captured in video, they focus on broader personal and social events or on the user's interaction with the image / video content that relates to such events. The latter interaction can provide useful cues for the event-based organization of the media items. In Concept-based image clustering and summarization of eventrelated image collections, the authors deal with the problem of summarizing image collections that correspond to a single event. To this end, they propose using trained visual concept detectors in combination with clustering methods. In Sentiment Flow for Video Interestingness Prediction, mthe authors develop a method for predicting how interesting a video is, employing a mid-level sentiment representation for the video content. In User Emotion Sensing in Search Process based on Chromatic Sensation, the authors propose a model for sensing the user's emotions by examining the colors (of photos, icons etc.) browsed or selected by the user. In Investigating Human Factors in Forgery Detection Process, the authors deal with the difficult problem of detecting if an image is forged or not (which is a question that often arises when an image of an event such as a natural disaster or a war incident surfaces on the Web). They conduct a subjective nevaluation to investigate human factors that are associated with this problem. Finally, in a position paper titled On the Personalization of Event-based Systems, the authors will describe their position about personalization as a paradigm shift and discuss its relation to the event-based processing and organization of signals and digital content.