Athanasia Zlatintsi

Multimodal Saliency and Fusion for Movie Summarization Based on Aural, Visual, and Textual Attention

Multimodal streams of sensory information are naturally parsed and integrated by humans using signal-level feature extraction and higher level cognitive processes. Detection of attention-invoking audiovisual segments is formulated in this work on the basis of saliency models for the audio, visual, and textual information conveyed in a video stream. Aural or auditory saliency is assessed by cues that quantify multifrequency waveform modulations, extracted through nonlinear operators and energy tracking. Visual saliency is measured through a spatiotemporal attention model driven by intensity, color, and orientation. Textual or linguistic saliency is extracted from part-of-speech tagging on the subtitles information available with most movie distributions. The individual saliency streams, obtained from modality-depended cues, are integrated in a multimodal saliency curve, modeling the time-varying perceptual importance of the composite video stream and signifying prevailing sensory events. The multimodal saliency representation forms the basis of a generic, bottom-up video summarization algorithm. Different fusion schemes are evaluated on a movie database of multimodal saliency annotations with comparative results provided across modalities. The produced summaries, based on low-level features and content-independent fusion and selection, are of subjectively high aesthetic and informative quality.

A supervised approach to movie emotion tracking

In this paper, we present experiments on continuous time, continuous scale affective movie content recognition (emotion tracking). A major obstacle for emotion research has been the lack of appropriately annotated databases, limiting the potential for supervised algorithms. To that end we develop and present a database of movie affect, annotated in continuous time, on a continuous valence-arousal scale. Supervised learning methods are proposed to model the continuous affective response using hidden Markov Models (independent) in each dimension. These models classify each video frame into one of seven discrete categories (in each dimension); the discrete-valued curves are then converted to continuous values via spline interpolation. A variety of audio-visual features are investigated and an optimal feature set is selected. The potential of the method is experimentally verified on twelve 30-minute movie clips with good precision at a macroscopic level.

Video event detection and summarization using audio, visual and text saliency

Detection of perceptually important video events is formulated here on the basis of saliency models for the audio, visual and textual information conveyed in a video stream. Audio saliency is assessed by cues that quantify multifrequency waveform modulations, extracted through nonlinear operators and energy tracking. Visual saliency is measured through a spatiotemporal attention model driven by intensity, color and motion. Text saliency is extracted from part-of-speech tagging on the subtitles information available with most movie distributions. The various modality curves are integrated in a single attention curve, where the presence of an event may be signified in one or multiple domains. This multimodal saliency curve is the basis of a bottom-up video summarization algorithm, that refines results from unimodal or audiovisual-based skimming. The algorithm performs favorably for video summarization in terms of informativeness and enjoyability.

Movie summarization based on audiovisual saliency detection

Based on perceptual and computational attention modeling studies, we formulate measures of saliency for an audiovisual stream. Audio saliency is captured by signal modulations and related multi-frequency band features, extracted through nonlinear operators and energy tracking. Visual saliency is measured by means of a spatiotemporal attention model driven by various feature cues (intensity, color, motion). Audio and video curves are integrated in a single attention curve, where events may be enhanced, suppressed or vanished. The presence of salient events is signified on this audiovisual curve by geometrical features such as local extrema, sharp transition points and level sets. An audiovisual saliency-based movie summarization algorithm is proposed and evaluated. The algorithm is shown to perform very well in terms of summary informativeness and enjoyability for movie clips of various genres.

Multiscale Fractal Analysis of Musical Instrument Signals With Application to Recognition

In this paper, we explore nonlinear methods, inspired by the fractal theory for the analysis of the structure of music signals at multiple time scales, which is of importance both for their modeling and for their automatic computer-based recognition. We propose the multiscale fractal dimension (MFD) profile as a short-time descriptor, useful to quantify the multiscale complexity and fragmentation of the different states of the music waveform. We have experimentally found that this descriptor can discriminate several aspects among different music instruments, which is verified by further analysis on synthesized sinusoidal signals. We compare the descriptiveness of our features against that of Mel frequency cepstral coefficients (MFCCs), using both static and dynamic classifiers such as Gaussian mixture models (GMMs) and hidden Markov models (HMMs). The method and features proposed in this paper appear to be promising for music signal analysis, due to their capability for multiscale analysis of the signals and their applicability in recognition, as they accomplish an error reduction of up to 32%. These results are quite interesting and render the descriptor of direct applicability in large-scale music classification tasks.

PREDICTING AUDIO-VISUAL SALIENT EVENTS BASED ON VISUAL, AUDIO AND TEXT MODALITIES FOR MOVIE SUMMARIZATION

In this paper, we present a new and improved synergistic approach to the problem of audio-visual salient event detection and movie summarization based on visual, audio and text modalities. Spatio-temporal visual saliency is estimated through a perceptually inspired frontend based on 3D (space, time) Gabor filters and frame-wise features are extracted from the saliency volumes. For the auditory salient event detection we extract features based on Teager-Kaiser Energy Operator, while text analysis incorporates part-of-speech tagging and affective modeling of single words on the movie subtitles. For the evaluation of the proposed system, we employ an elementary and non-parametric classification technique like KNN. Detection results are reported on the MovSum database, using objective evaluations against ground-truth denoting the perceptually salient events, and human evaluations of the movie summaries. Our evaluation verifies the appropriateness of the proposed methods compared to our baseline system. Finally, our newly proposed summarization algorithm produces summaries that consist of salient and meaningful events, also improving the comprehension of the semantics.

Audio salient event detection and summarization using audio and text modalities

This paper investigates the problem of audio event detection and summarization, building on previous work [1,2] on the detection of perceptually important audio events based on saliency models. We take a synergistic approach to audio summarization where saliency computation of audio streams is assisted by using the text modality as well. Auditory saliency is assessed by auditory and perceptual cues such as Teager energy, loudness and roughness; all known to correlate with attention and human hearing. Text analysis incorporates part-of-speech tagging and affective modeling. A computational method for the automatic correction of the boundaries of the selected audio events is applied creating summaries that consist not only of salient but also meaningful and semantically coherent events. A non-parametric classification technique is employed and results are reported on the MovSum movie database using objective evaluations against ground-truth designating the auditory and semantically salient events.

Audio-Based Distributional Representations of Meaning Using a Fusion of Feature Encodings.

Recently a “Bag-of-Audio-Words” approach was proposed [1] for the combination of lexical features with audio clips in a multimodal semantic representation, i.e., an Audio Distributional Semantic Model (ADSM). An important step towards the creation of ADSMs is the estimation of the semantic distance between clips in the acoustic space, which is especially challenging given the diversity of audio collections. In this work, we investigate the use of different feature encodings in order to address this challenge following a two-step approach. First, an audio clip is categorized with respect to three classes, namely, music, speech and other. Next, the feature encodings are fused according to the posterior probabilities estimated in the previous step. Using a collection of audio clips annotated with tags we derive a mapping between words and audio clips. Based on this mapping and the proposed audio semantic distance, we construct an ADSM model in order to compute the distance between words (lexical semantic similarity task). The proposed model is shown to significantly outperform (23.6% relative improvement in correlation coefficient) the state-of-the-art results reported in the literature.

Quality evaluation of computational models for movie summarization

In this paper we present a movie summarization system and we investigate what composes high quality movie summaries in terms of user experience evaluation. We propose state-of-the-art audio, visual and text techniques for the detection of perceptually salient events from movies. The evaluation of such computational models is usually based on the comparison of the similarity between the system-detected events and some ground-truth data. For this reason, we have developed the MovSum movie database, which includes sensory and semantic saliency annotation as well as cross-media relations, for objective evaluations. The automatically produced movie summaries were qualitatively evaluated, in an extensive human evaluation, in terms of informativeness and enjoyability accomplishing very high ratings up to 80% and 90%, respectively, which verifies the appropriateness of the proposed methods.

Social Human-Robot Interaction for the Elderly: Two Real-life Use Cases

We explore new aspects on assistive living via smart social human-robot interaction (HRI) involving automatic recognition of multimodal gestures and speech in a natural interface, providing social features in HRI. We discuss a whole framework of resources, including datasets and tools, briefly shown in two real-life use cases for elderly subjects: a multimodal interface of an assistive robotic rollator and an assistive bathing robot. We discuss these domain specific tasks, and open source tools, which can be used to build such HRI systems, as well as indicative results. Sharing such resources can open new perspectives in assistive HRI.