Learning Temporal Regularity in Video Sequences Hasan, Mahmudul; Jonghyun Choi; Neumann, Jan ...
2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),
06/2016
Conference Proceeding
Open access
Perceiving meaningful activities in a long video sequence is a challenging problem due to ambiguous definition of 'meaningfulness' as well as clutters in the scene. We approach this problem by ...learning a generative model for regular motion patterns (termed as regularity) using multiple sources with very limited supervision. Specifically, we propose two methods that are built upon the autoencoders for their ability to work with little to no supervision. We first leverage the conventional handcrafted spatio-temporal local features and learn a fully connected autoencoder on them. Second, we build a fully convolutional feed-forward autoencoder to learn both the local features and the classifiers as an end-to-end learning framework. Our model can capture the regularities from multiple datasets. We evaluate our methods in both qualitative and quantitative ways - showing the learned regularity of videos in various aspects and demonstrating competitive performance on anomaly detection datasets as an application.
We introduce the task of Multi-Modal Machine Comprehension (M3C), which aims at answering multimodal questions given a context of text, diagrams and images. We present the Textbook Question Answering ...(TQA) dataset that includes 1,076 lessons and 26,260 multi-modal questions, taken from middle school science curricula. Our analysis shows that a significant portion of questions require complex parsing of the text and the diagrams and reasoning, indicating that our dataset is more complex compared to previous machine comprehension and visual question answering datasets. We extend state-of-the-art methods for textual machine comprehension and visual question answering to the TQA dataset. Our experiments show that these models do not perform well on TQA. The presented dataset opens new challenges for research in question answering and reasoning across multiple modalities.
Meso-porous Si-coated carbon nanotube (CNT) composite powders were prepared by combining a sol-gel method and the magnesiothermic reduction process. Meso-porous Si-coated CNT electrodes exhibit ...excellent cycle and rate performances as anodes in Li-ion batteries (LIBs), which can be attributed to the efficient accommodation of volume change from meso-porous Si structure and the enhanced electrical conductivity from CNT core. This simple synthesis and subsequent reduction process provide a scalable route for the large-scale production of Si-C composite nanostructures, which can be utilized in a variety of applications, such as in photocatalysis, photoelectrochemical cells (PECs), and LIBs.
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
Plasma-activated water (PAW) is generated by treating water with cold atmospheric plasma (CAP) using controllable parameters, such as plasma-forming voltage, carrier gas, temperature, pulses, or ...frequency as required. PAW is reported to have lower pH, higher conductivity, and higher oxygen reduction potential when compared with untreated water due to the presence of reactive species. PAW has received significant attention from researchers over the last decade due to its non-thermal and non-toxic mode of action especially for bacterial inactivation. The objective of the current review is to develop a summary of the effect of PAW on bacterial strains in foods as well as model systems such as buffers, with a specific focus on fruit and vegetables. The review elaborated the properties of PAW, the effect of various treatment parameters on its efficiency in bacterial inactivation along with its usage as a standalone technology as well as a hurdle approach with mild thermal treatments. A section highlighting different models that can be employed to generate PAW alongside a direct comparison of the PAW characteristics on the inactivation potential and the existing research gaps are also included. The mechanism of action of PAW on the bacterial cells and any reported effects on the sensory qualities and shelf life of food has been evaluated. Based on the literature, it can be concluded that PAW offers a significant potential as a non-chemical and non-thermal intervention for bacterial inactivation, especially on food. However, the applicability and usage of PAW depend on the effect of environmental and bacterial strain-based conditions and cost-effectiveness.
Joint Image Clustering and Labeling by Matrix Factorization Seunghoon Hong; Jonghyun Choi; Feyereisl, Jan ...
IEEE transactions on pattern analysis and machine intelligence,
2016-July-1, 2016-07-00, 2016-7-1, 20160701, Volume:
38, Issue:
7
Journal Article
Peer reviewed
We propose a novel algorithm to cluster and annotate a set of input images jointly, where the images are clustered into several discriminative groups and each group is identified with representative ...labels automatically. For these purposes, each input image is first represented by a distribution of candidate labels based on its similarity to images in a labeled reference image database. A set of these label-based representations are then refined collectively through a non-negative matrix factorization with sparsity and orthogonality constraints; the refined representations are employed to cluster and annotate the input images jointly. The proposed approach demonstrates performance improvements in image clustering over existing techniques, and illustrates competitive image labeling accuracy in both quantitative and qualitative evaluation. In addition, we extend our joint clustering and labeling framework to solving the weakly-supervised image classification problem and obtain promising results.
Wireless sensor tags in flexible formats have numerous applications; some are commercially available for specific target applications. However, most of these wireless sensor tags have been used for ...single-sensing applications. In this study, we designed a printed circuit board (PCB) module (13 mm × 13 mm) for near-field communication-enabled sensor tags with both electrical resistance and capacitance read-out channels that enables dual-channel sensing. As part of the wireless sensor tag, a square antenna pattern was printed directly on a flexible poly(ethylene terephthalate) (PET) substrate and integrated into the PCB module to demonstrate a dual-channel temperature and ethylene gas sensor. The temperature and ethylene sensors were printed using a positive temperature coefficient ink and a tin oxide (SnO
) nanoparticle ink, respectively. With dual sensing capabilities, this type of sensor tag can be used in smart packaging for the quality monitoring of fresh produce (e.g., bananas) by tracking temperature and ethylene concentration in the storage/transport environment.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, PNG, SAZU, UL, UM, UPUK
Tin phosphide (Sn4P3) has emerged as an anode for sodium ion batteries (SIBs) due to its high reversible capacity and low redox potential. Sn4P3 shows a synergistic Na-storage reaction to form ...Na15Sn4 and Na3P, but suffers from large volume expansion and Sn aggregation during the Na+ insertion–extraction resulting in poor cycle stability. Sn4P3 has also been considered a promising anode material for lithium ion batteries (LIBs), but very limited studies have been performed. Herein, core–shell Sn4P3–C (carbon) composite nanospheres are fabricated by carbonization/reduction and phosphorization of SnO2–GCP (glucose-derived, carbon-rich polysaccharide) nanospheres. The size of Sn4P3–C nanospheres is controlled to optimize their electrochemical performance as long-term stable anodes for SIBs and LIBs. Among them, the 140 nm-sized Sn4P3–C nanosphere electrode exhibits high reversible capacity, high rate capability, and ultra-long cycle stability as an anode for both SIBs and LIBs, delivering a high capacity of 420 mA h g−1 after 2000 cycles (SIBs) and 440 mA h g−1 after 500 cycles (LIBs) at a high current density of 2000 mA g−1. Hence, the Sn4P3–C nanospheres can be considered as a promising anode material for next generation SIBs and LIBs.
Single wall carbon nanotube (SWCNT) and reduced graphene oxide (RGO) reinforced alumina ceramic composites were fabricated, and their microstructure, interface between matrix and reinforcement, ...electrical, mechanical, and thermal properties were comparatively investigated. The homogeneous dispersion of SWCNT and RGO and nearly full densification (> 96% of theoretical density) were achieved by dimethylformamide (DMF) solvent and spark plasma sintering (SPS), respectively. The chemical bonding between alumina and reinforcements was examined by high resolution transmission electron microscopy (HRTEM) with electron energy loss spectroscopy (EELS) and X-ray photoelectron spectroscopy (XPS). Elastic modulus and hardness of Al2O3 composites decreased, but fracture toughness and flexural strength increased with addition of SWCNT and RGO. SWCNT outperformed RGO in flexural strength whereas RGO was better than SWCNT to improve the fracture toughness. SWCNT was more effective for high electrical conductivity and low percolation threshold than RGO. The thermal conductivity of both composites decreased compared to monolithic Al2O3 possibly due to the reduced phonon mean free path by defects and carbon contamination.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK, ZRSKP
Fine-grained classification involves distinguishing between similar sub-categories based on subtle differences in highly localized regions, therefore, accurate localization of discriminative regions ...remains a major challenge. We describe a patch-based framework to address this problem. We introduce triplets of patches with geometric constraints to improve the accuracy of patch localization, and automatically mine discriminative geometrically-constrained triplets for classification. The resulting approach only requires object bounding boxes. Its effectiveness is demonstrated using four publicly available fine-grained datasets, on which it outperforms or achieves comparable performance to the state-of-the-art in classification.
V4P7 nanoparticles were synthesized via high-energy mechanical milling and their electrochemical properties as an anode for sodium-ion batteries were studied and compared with those of VO2(B)/Na and ...V4P7/Li cells, focusing on the electrochemical reaction mechanism and cycle performance. The V4P7 showed excellent cycling behavior even without any conductive material.