Affective Content Analysis

For individuals undergoing psychiatric rehabilitation, there is an urgent need to establish an objective biomarker of brain activity without invasive or behavioral restrictions. To investigate an objective index for potential psychiatric rehabilitation assessment and intervention strategies in the future, we tried to detect the "Berger effect", the alpha EEG reactivity to eyes opening, and to verify the suitability of a sheet-type electroencephalography (EEG) device, HARU-1. We measured EEG of 20 healthy young to middle-aged subjects using HARU-1 from the forehead area during resting closed-eye and open-eye conditions. Fast Fourier transform and timefrequency analysis were then performed on the data from all subjects in the resting closed-eye and resting open-eye conditions. Statistical analysis found that the reference electrode derivation method showed a significant increase in alpha waves and a significant decrease in beta and gamma waves in the three forehead channels of Fp1, Fp2 and AFz during the resting closed eye condition compared to the resting open eye condition. In the bipolar derivation method, all Fp1-AFz, Fp2-Fp1 and AFz-Fp2 showed a significant decrease in the b and g bands at approximately 13-50 Hz. These results demonstrated that HARU-1 accurately measured the "Berger effect" in the forehead region, suggesting that HARU-1 may be useful in detecting objective biomarkers of brain activity during psychiatric rehabilitation.

More and more popular, Smart TVs and set-top boxes open new ways for richer experiences in our living rooms. But to offer richer and novel functionalities, a better understanding of the multimedia content is crucial. If many works try to automatically annotate videos at object level, or classify them, we think that investigating the emotions through the use of digital analysis and processing techniques will allow great TV experience improvements. With our work, we propose a temporal saliency detection approach capable of defining the most exciting parts of a video that will be of the most interest to the users. To identify the most interesting events, without performing their classification (in order to be independent from the video domain), we compute a time series of arousal (excitement level of the content), based on audio-visual features. Our goal is to merge this preliminary work with user emotions analysis, in order to create a multimodal system, allowing to bridge the gap between users' needs and multimedia contents.

More and more popular, Smart TVs and set-top boxes open new ways for richer experiences in our living rooms. But to offer richer and novel functionalities, a better understanding of the multimedia content is crucial. If many works try to automatically annotate videos at object level, or classify them, we think that investigating the emotions through the use of digital analysis and processing techniques will allow great TV experience improvements. With our work, we propose a temporal saliency detection approach capable of defining the most exciting parts of a video that will be of the most interest to the users. To identify the most interesting events, without performing their classification (in order to be independent from the video domain), we compute a time series of arousal (excitement level of the content), based on audiovisual features. Our goal is to merge this preliminary work with user emotions analysis, in order to create a multimodal system, allowing to bridge the gap between users' needs and multimedia contents.

Texture is one of the important characteristics used in identifying objects (or) regions of interest in an image. This can be identified by aerial or satellite photographs, biomedical images and other types of images [1]. In the field of computer vision, texture classification is an important task. Texture classification is used in different pattern recognition application. It retains feature-liked appearance. This paper examines, analyzing the feature extraction towards texture and non-texture classification. In this paper we present texture classification and the feature extraction methods used in the research. Different extraction methods were introduced and used for texture classification problems.

Different drivers' states and emotions can affect negatively the driving performance. Recent advances in affective computing now give the opportunity to measure the users' state or emotions using various sources of data such as physiological signals or voice samples. Conveying biofeedback in the car could help to make roads safer and improve users' health and mental state during a ride in an autonomous car. This workshop aims at selecting the drivers' hazardous states and emotions that are crucial to be assessed, as well as how to convey the appropriate biofeedback to the driver, using multimodal interaction in the car. CCS CONCEPTS • Human-centered computing → HCI theory, concepts and models; Visualization design and evaluation methods.

More and more popular, Smart TVs and set-top boxes open new ways for richer experiences in our living rooms. But to offer richer and novel functionalities, a better understanding of the multimedia content is crucial. If many works try to automatically annotate videos at object level, or classify them, we think that investigating the emotions through the use of digital analysis and processing techniques will allow great TV experience improvements. With our work, we propose a temporal saliency detection approach capable of defining the most exciting parts of a video that will be of the most interest to the users. To identify the most interesting events, without performing their classification (in order to be independent from the video domain), we compute a time series of arousal (excitement level of the content), based on audiovisual features. Our goal is to merge this preliminary work with user emotions analysis, in order to create a multimodal system, allowing to bridge the gap between users' needs and multimedia contents.

Automatic semantic annotation of videos is a crucial to the success of video search and summarisation based on content semantics. In contrast to broadcast news and sports, automatic semantic annotation for non-commercial videos generated by ordinary people suffers from lack of semantic data like subtitles and webcast text. It is, however, impracticable to expect that most video shooters or owners of the non-commercial contents do semantic annotation of their videos by using annotation tools manually before video dissemination. This paper proposes a video semantic annotation system that automatically analyzes and annotates a video element with the user attention state. The attention state includes the attention target, attention degree and emotional states by using gaze and Electroencephalography data from a user watching the video. To show the benefits achieved by the proposed system, the paper describes a promising application scenario of video summarisation using semantic annotati...

In modern world, hearing a song or seeing a video has become an imperative entertainment to people. Music Video Content (MVC) must be retrieved based on emotional information on human presence of mind. Many researches focus the study of relationship between videos and users’ induced physiological and psychological responses. The existing system performs the emotion analysis by using single-trial classification with arousal, valence and liking using features extracted from the electroencephalogram (EEG) and peripheral physiological signals and MCA (Multimedia Content Analysis) modalities. It uses semi-automatic stimuli selection method using affective tags, which was validated by an analysis of the ratings given by the participants. But, it has limitations while considering the signal noise, physiological differences among individuals, and limited quality of self-assessments. To overcome these limitations, it is necessary to develop a new technique for effective MVC model. In the pro...

More and more popular, Smart TVs and set-top boxes open new ways for richer experiences in our living rooms. But to offer richer and novel functionalities, a better understanding of the multimedia content is crucial. If many works try to automatically annotate videos at object level, or classify them, we think that investigating the emotions through the use of digital analysis and processing techniques will allow great TV experience improvements. With our work, we propose a temporal saliency detection approach capable of defining the most exciting parts of a video that will be of the most interest to the users. To identify the most interesting events, without performing their classification (in order to be independent from the video domain), we compute a time series of arousal (excitement level of the content), based on audiovisual features. Our goal is to merge this preliminary work with user emotions analysis, in order to create a multi-modal system, allowing to bridge the gap bet...

ABSTRACT If new televisions are bringing more and more features in our living rooms, they are still lacking an important axis of the TV user experience: emotions. And this is particularly true for visually and hearing impaired people. Therefore, we want to find innovative ways to bring them emotions that are usually conveyed through sound and images, by taking advantage of arousal modeling of films. Such empathic TV experiences will give them the possibility to fully enjoy visual media.