For high-level feature extraction, we submitted 4 automatic runs: Fudan.Global: this run is based on global features of keyframes. Fudan.Local: this run is based on local features of keyframes. Fudan.Rerank: this run is based on local features and spatial information of keyframes. Fudan.Fusion: this run is based on the fusion of global and local features of keyframes. Focus of our system was on...

Novel and simplified methods for determining low-level states of student behavior and predicting affective states enable tutors to better respond to students. The Many Eyes Word Tree graphics is used to understand and analyze sequential patterns of student states, categorizing raw quantitative indicators into a limited number of discrete sates. Used in combination with sensor predictors, we dem...

In this paper a general and e cient approach for representing and classifying image sequences by Hid den Markov Models HMMs is presented A consis tent modeling of spatial and temporal information is achieved by extracting di erent low level image fea tures These implicitly convert the image intensities into probability density values while preserving the ge ometry of the image The resulting so ...

We take part in the short text conversation task at NTCIR-12. We employ a semantic-based retrieval method to tackle this problem, by calculating textual similarity between posts and comments. Our method applies a rich-feature model to match post-comment pairs, by using semantic, grammar, n-gram and string features to extract high-level semantic meanings of text.

We participated in the high-level feature extraction task in TRECVID 2007. This paper describes the details of our system for the task. For feature extraction, we propose an EMD-based bag-of-feature method to exploit visual/spatial information, and utilize WordNet to expand semantic meanings of text to boost up the generalization of detectors. We also explore audio features and extract the moti...

Due to the short duration and low intensity of micro-expressions, the recognition of micro-expression is still a challenging problem. In this paper, we develop a novel multi-task mid-level feature learning method to enhance the discrimination ability of extracted low-level features by learning a set of class-specific feature mappings, which would be used for generating our mid-level feature rep...

Low-level features tend to achieve unsatisfactory retrieval results in remote sensing image retrieval community because of the existence of semantic gap. In order to improve retrieval precision, visual attention model is used to extract salient objects from image according to their saliency. Then color and texture features are extracted from salient objects and regarded as feature vectors for i...

Road, railway and track networks are important features of satellite imagery. Automated detection of tracks in mountainous terrain using satellite imagery is a difficult task. The track and the shadow signature of the mountainous terrain are almost same. The presence of ridges and snow also creates a lot of ambiguity. Moreover due to poor sunlight condition in the mountainous terrain the images...

Bilkent University Multimedia Database Group (BILMDG) participated in two tasks at TRECVID 2008: content-based copy detection (CBCD) and high-level feature extraction (FE). Mostly MPEG-7 [1] visual features, which are also used as low-level features in our MPEG-7 compliant video database management system, are extracted for these tasks. This paper discusses our approaches in each task.

This paper aims to show that by using low level feature extraction, motion and object identifying and tracking methods, features can be extracted and indexed for eYcient and eVective retrieval for video; such as an awards ceremony video. Video scene/shot analysis and key frame extraction are used as a foundation to identify objects in video and be able to Wnd spatial relationships within the vi...