In this paper we discuss several forms of spatiotemporal convolutions for video analysis and study their effects on action recognition. Our motivation stems from the observation that 2D CNNs applied to individual frames of the video have remained solid performers in action recognition. In this work we empirically demonstrate the accuracy advantages of 3D CNNs over 2D CNNs within the framework of residual learning. Furthermore, we show that factorizing the 3D convolutional filters into separate spatial and temporal components yields significantly gains in accuracy. Our empirical study leads to the design of a new spatiotemporal convolutional block "R(2+1)D" which produces CNNs that achieve results comparable or superior to the state-of-theart on Sports-1M, Kinetics, UCF101, and HMDB51.
This three-phase investigation used focus groups and a survey to identify factors that perceived by speech language pathologists as being related to long-term success versus inappropriate abandonment of augmentative and alternative communication (AAC) systems. Factors deemed most important by six focus groups were included in a 106-question survey that was returned by 275 ASHA Special Interest Division #12 (AAC) members. Factor analysis indicated the constructs of Support, Attitude, and System characteristics and Fit as most important to the long-term success of AAC systems. The constructs of Not Maintaining/Adjusting the System, Attitude, Lack of Training, Lack of Support, and Poor Fit were most often related to inappropriate abandonment of AAC systems. Systematic implementation of intervention targeting the constructs is recommended.
Learning image representations with ConvNets by pretraining on ImageNet has proven useful across many visual understanding tasks including object detection, semantic segmentation, and image captioning. Although any image representation can be applied to video frames, a dedicated spatiotemporal representation is still vital in order to incorporate motion patterns that cannot be captured by appearance based models alone. This paper presents an empirical ConvNet architecture search for spatiotemporal feature learning, culminating in a deep 3-dimensional (3D) Residual ConvNet. Our proposed architecture outperforms C3D by a good margin on Sports-1M, UCF101, HMDB51, THU-MOS14, and ASLAN while being 2 times faster at inference time, 2 times smaller in model size, and having a more compact representation.
Barren Island (India) is a relatively little studied, little known active volcano in the Andaman Sea, and the northernmost active volcano of the great Indonesian arc. The volcano is built of prehistoric (possibly late Pleistocene) lava flows (dominantly basalt and basaltic andesite, with minor andesite) intercalated with volcaniclastic deposits (tuff breccias, and ash beds deposited by pyroclastic falls and surges), which are exposed along a roughly circular caldera wall. There are indications of a complete phreatomagmatic tephra ring around the exposed base of the volcano. A polygenetic cinder cone has existed at the centre of the caldera and produced basalt-basaltic andesite aa and blocky aa lava flows, as well as tephra, during historic eruptions (1787-1832) and three recent eruptions (1991,. The recent aa flows include a toothpaste aa flow, with tilted and overturned crustal slabs carried atop an aa core, as well as locally developed tumuli-like elliptical uplifts having corrugated crusts. Based on various evidence we infer that it belongs to either the 1991 or the 1994-95 eruptions. The volcano has recently (2008) begun yet another eruption, so far only of tephra. We make significantly different interpretations of several features of the volcano than previous workers. This study of the volcanology and eruptive styles of the Barren Island volcano lays the ground for detailed geochemical-isotopic and petrogenetic work, and provides clues to what the volcano can be expected to do in the future.
The Vindhyan Supergroup of India is one of the largest and thickest sedimentary successions of the world. Deposited in an intra-cratonic basin, it is composed mostly of shallow marine deposits. It is believed to have recorded a substantial portion of Proterozoic time and therefore, likely to contain valuable information on the evolution of the atmosphere, climate, and life on our planet. It also contains some of the most disputed fossils of earliest animal life. Despite their importance, the absolute age of these rocks had remained unknown until recently. In this work I evaluate all the recent chronological information and discuss their implications. From the present findings it appears that the issues surrounding the age of the Lower Vindhyans in the Son valley are now resolved, whereas problems with the age of the Upper Vindhyans and that with the stratigraphic correlations remain to be answered.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.