Abstract. In this paper, we present a framework for estimating what portions of videos are most discriminative for the task of action recognition. We explore the impact of the temporal cropping of training videos on the overall accuracy of an action recognition system, and we formalize what makes a

We propose temporal extent, a concept that can be used to produce summary or abstracted information about the values of temporal attributes in a database relation or class of objects. In its simplest form, the temporal extent is a one-dimensional interval representing the minimum and maximum time

is interpreted kinematically by abrupt termination of rupture expansion at the end of the fault. We use the stopping phase to determine total source extent in space and time, as well as the mean rupture velocity. The results of this first stage of investigation are subsequently used for modeling the rupture

The goal of this study is to investigate the temporal extent of nasalization in French oral vowels followed by nasal consonants. The extent of nasalization is obtained by measuring the acoustic signal at the nostrils with an appropriate microphone for 14 French speakers. Total duration

and validate such sampling guidelines. Applying geostatistical analysis, we quantified the spatio-temporal variability of parameters, which characterize the physico-chemical substratum conditions in the hyporheic zone. We investigated eight stream reaches in six small streams that are typical for the majority

Abstract not found

Recorded speech poses unusual challenges for the design of interactive end-user search systems. Automatic speech recognition is sufficiently accurate to support the automated components of interactive search systems in some applications, but finding useful recordings among those nominated by the system can be difficult because listening to audio is time consuming and because recognition errors and speech disfluencies make it difficult to mitigate that effect by skimming automatic transcripts. Support for rapid browsing based on supervised learning for automatic classification has shown promise, however, and a segment-then-label framework has emerged as the dominant paradigm for applying that technique to news broadcasts. This paper argues for a more general framework, which we call an activation matrix, that provides a flexible representation for the mapping between labels and time. Three approaches to the generation of activation matrices are briefly described, with the main focus of the paper then being the use of activation matrices to support search and selection in interactive systems.

Recorded speech poses unusual challenges for the design of interactive end-user search systems. Automatic speech recognition is sufficiently accurate to support the automated components of interactive search systems in some applications. Recognizing useful recordings among those nominated by the system is difficult, however, because listening to audio is time consuming and because recognition errors and speech disfluencies make it difficult to mitigate this time factor by skimming automatic transcripts. Support for the browsing process based on supervised learning for automatic classification has shown promise, however, and a segment-then-label framework has emerged as the dominant paradigm for applying that technique to news broadcasts. This paper argues for a more general framework, which we call activation matrices, that provide a flexible representation for the mapping between labels and time. Three approaches to the generation of activation matrices could be generated are briefly described, with the main focus of the paper being the use of activation matrices to support search and selection in interactive systems.

This paper describes a simple approach for modifying an input-output (or queueing) diagram to measure the time and distance spent by vehicles in a queue in a much simpler and self-serving manner than using a time-space diagram. The graphical technique requires construction of a curve depicting the cumulative number of vehicles to have reached the back of the queue as a function of time, but as shown herein, the technique can be easily automated with a spreadsheet.

Abstract.We assessed the consistency across space and time of spatially explicit models of forest presence and biomass in southern Missouri, USA, for adjacent, partially overlapping satellite image Path/Rows, and for coincident satellite images from the same Path/Row acquired in different years. Such consistency in satellite image-based classification and estimation is critical to national and continental monitoring programs that depend upon processed satellite imagery, such as the North American Forest Dynamics Program. We tested the interchangeability of particular image acquisitions across time and space in the context of modeling forest biomass and forest presence with a non-parametric Random Forests-based approach. Validation at independent USA national forest inventory plots suggested statistically consistent model accuracy, even when the images used to apply the models were acquired in different years or in differ-ent image frames from the images used to build the models. For mapping projects using near-anniversary date imagery and employing careful radiometric correction, advantages of image interchangeability include the ability to build models with more ground data by combining adjacent image frames and the ability to apply models of assessed accuracy to early satellite images for which no corresponding field data may be available.