The Experts below are selected from a list of 34104 Experts worldwide ranked by ideXlab platform
P Lopez - One of the best experts on this subject based on the ideXlab platform.
-
deep learning based multiple object visual tracking on embedded system for iot and mobile edge computing applications
IEEE Internet of Things Journal, 2019Co-Authors: B Blancofilgueira, Daniel Garcialesta, Mauro Fernandezsanjurjo, V M Brea, P LopezAbstract:Compute and memory demands of state-of-the-art deep learning methods are still a shortcoming that must be addressed to make them useful at Internet of Things (IoT) end-nodes. In particular, recent results depict a hopeful prospect for image processing using convolutional neural networks, CNNs, but the gap between software and hardware implementations is already considerable for IoT and mobile edge computing applications due to their high power consumption. This proposal performs low-power and real time deep learning-based multiple object visual tracking implemented on an NVIDIA Jetson TX2 development kit. It includes a camera and wireless connection capability and it is battery powered for mobile and outdoor applications. A collection of Representative Sequences captured with the on-board camera, dETRUSC video dataset, is used to exemplify the performance of the proposed algorithm and to facilitate benchmarking. The results in terms of power consumption and frame rate demonstrate the feasibility of deep learning algorithms on embedded platforms although more effort in the joint algorithm and hardware design of CNNs is needed.
-
deep learning based multiple object visual tracking on embedded system for iot and mobile edge computing applications
arXiv: Computer Vision and Pattern Recognition, 2018Co-Authors: B Blancofilgueira, Daniel Garcialesta, Mauro Fernandezsanjurjo, V M Brea, P LopezAbstract:Compute and memory demands of state-of-the-art deep learning methods are still a shortcoming that must be addressed to make them useful at IoT end-nodes. In particular, recent results depict a hopeful prospect for image processing using Convolutional Neural Netwoks, CNNs, but the gap between software and hardware implementations is already considerable for IoT and mobile edge computing applications due to their high power consumption. This proposal performs low-power and real time deep learning-based multiple object visual tracking implemented on an NVIDIA Jetson TX2 development kit. It includes a camera and wireless connection capability and it is battery powered for mobile and outdoor applications. A collection of Representative Sequences captured with the on-board camera, dETRUSC video dataset, is used to exemplify the performance of the proposed algorithm and to facilitate benchmarking. The results in terms of power consumption and frame rate demonstrate the feasibility of deep learning algorithms on embedded platforms although more effort to joint algorithm and hardware design of CNNs is needed.
B Blancofilgueira - One of the best experts on this subject based on the ideXlab platform.
-
deep learning based multiple object visual tracking on embedded system for iot and mobile edge computing applications
IEEE Internet of Things Journal, 2019Co-Authors: B Blancofilgueira, Daniel Garcialesta, Mauro Fernandezsanjurjo, V M Brea, P LopezAbstract:Compute and memory demands of state-of-the-art deep learning methods are still a shortcoming that must be addressed to make them useful at Internet of Things (IoT) end-nodes. In particular, recent results depict a hopeful prospect for image processing using convolutional neural networks, CNNs, but the gap between software and hardware implementations is already considerable for IoT and mobile edge computing applications due to their high power consumption. This proposal performs low-power and real time deep learning-based multiple object visual tracking implemented on an NVIDIA Jetson TX2 development kit. It includes a camera and wireless connection capability and it is battery powered for mobile and outdoor applications. A collection of Representative Sequences captured with the on-board camera, dETRUSC video dataset, is used to exemplify the performance of the proposed algorithm and to facilitate benchmarking. The results in terms of power consumption and frame rate demonstrate the feasibility of deep learning algorithms on embedded platforms although more effort in the joint algorithm and hardware design of CNNs is needed.
-
deep learning based multiple object visual tracking on embedded system for iot and mobile edge computing applications
arXiv: Computer Vision and Pattern Recognition, 2018Co-Authors: B Blancofilgueira, Daniel Garcialesta, Mauro Fernandezsanjurjo, V M Brea, P LopezAbstract:Compute and memory demands of state-of-the-art deep learning methods are still a shortcoming that must be addressed to make them useful at IoT end-nodes. In particular, recent results depict a hopeful prospect for image processing using Convolutional Neural Netwoks, CNNs, but the gap between software and hardware implementations is already considerable for IoT and mobile edge computing applications due to their high power consumption. This proposal performs low-power and real time deep learning-based multiple object visual tracking implemented on an NVIDIA Jetson TX2 development kit. It includes a camera and wireless connection capability and it is battery powered for mobile and outdoor applications. A collection of Representative Sequences captured with the on-board camera, dETRUSC video dataset, is used to exemplify the performance of the proposed algorithm and to facilitate benchmarking. The results in terms of power consumption and frame rate demonstrate the feasibility of deep learning algorithms on embedded platforms although more effort to joint algorithm and hardware design of CNNs is needed.
Jin Yuanxiao - One of the best experts on this subject based on the ideXlab platform.
-
RESEARCH ARTICLE Fungal Endophytes of Alpinia officinarum Rhizomes: Insights on Diversity and Variation across Growth Years, Growth Sites, and the Inner Active Chemical Concentration
2016Co-Authors: Li Shubin, Huang Juan, Zhou Renchao, Xu Shiru, Jin YuanxiaoAbstract:In the present study, the terminal-restriction fragment length polymorphism (T-RFLP) technique, combined with the use of a clone library, was applied to assess the baseline diversity of fungal endophyte communities associated with rhizomes of Alpinia officinarum Hance, a medicinal plant with a long history of use. A total of 46 distinct T-RFLP fragment peaks were detected using HhaI or MspI mono-digestion-targeted, amplified fungal rDNA ITS Sequences from A. officinarum rhizomes. Cloning and sequencing of Representative Sequences resulted in the detection of members of 10 fungal genera: Pestalotiopsis, Sebacina, Penicillium, Marasmius, Fusarium, Exserohilum, Mycoleptodiscus, Colletotrichum, Meyerozyma, and Scopulariopsis. The T-RFLP profiles revealed an influence of growth year of the host plant on fungal endophyte communities in rhizomes of this plant species; whereas, the geographic location where A. officinarum was grown contributed to only limited variation in the fungal endophyte communities of the host tissue. Furthermore, non-metric multidimensional scaling (NMDS) analysis across all of th
-
fungal endophytes of alpinia officinarum rhizomes insights on diversity and variation across growth years growth sites and the inner active chemical concentration
PLOS ONE, 2014Co-Authors: Li Shubin, Huang Juan, Zhou Renchao, Xu Shiru, Jin YuanxiaoAbstract:In the present study, the terminal-restriction fragment length polymorphism (T-RFLP) technique, combined with the use of a clone library, was applied to assess the baseline diversity of fungal endophyte communities associated with rhizomes of Alpinia officinarum Hance, a medicinal plant with a long history of use. A total of 46 distinct T-RFLP fragment peaks were detected using HhaI or MspI mono-digestion-targeted, amplified fungal rDNA ITS Sequences from A. officinarum rhizomes. Cloning and sequencing of Representative Sequences resulted in the detection of members of 10 fungal genera: Pestalotiopsis, Sebacina, Penicillium, Marasmius, Fusarium, Exserohilum, Mycoleptodiscus, Colletotrichum, Meyerozyma, and Scopulariopsis. The T-RFLP profiles revealed an influence of growth year of the host plant on fungal endophyte communities in rhizomes of this plant species; whereas, the geographic location where A. officinarum was grown contributed to only limited variation in the fungal endophyte communities of the host tissue. Furthermore, non-metric multidimensional scaling (NMDS) analysis across all of the rhizome samples showed that the fungal endophyte community assemblages in the rhizome samples could be grouped according to the presence of two types of active indicator chemicals: total volatile oils and galangin. Our present results, for the first time, address a diverse fungal endophyte community is able to internally colonize the rhizome tissue of A. officinarum. The diversity of the fungal endophytes found in the A. officinarum rhizome appeared to be closely correlated with the accumulation of active chemicals in the host plant tissue. The present study also provides the first systematic overview of the fungal endophyte communities in plant rhizome tissue using a culture-independent method.
Thorsten Lumbsch - One of the best experts on this subject based on the ideXlab platform.
-
fungal specificity and selectivity for algae play a major role in determining lichen partnerships across diverse ecogeographic regions in the lichen forming family parmeliaceae ascomycota
Molecular Ecology, 2015Co-Authors: Steven D Leavitt, Ekaphan Kraichak, Matthew P Nelsen, Susanne Altermann, Pradeep K Divakar, David Alors, Theodore L Esslinger, Ana Crespo, Thorsten LumbschAbstract:Microbial symbionts are instrumental to the ecological and long-term evolutionary success of their hosts, and the central role of symbiotic interactions is increasingly recognized across the vast majority of life. Lichens provide an iconic group for investigating patterns in species interactions; however, relationships among lichen symbionts are often masked by uncertain species boundaries or an inability to reliably identify symbionts. The species-rich lichen-forming fungal family Parmeliaceae provides a diverse group for assessing patterns of interactions of algal symbionts, and our study addresses patterns of lichen symbiont interactions at the largest geographic and taxonomic scales attempted to date. We analysed a total of 2356 algal internal transcribed spacer (ITS) region Sequences collected from lichens representing ten mycobiont genera in Parmeliaceae, two genera in Lecanoraceae and 26 cultured Trebouxia strains. Algal ITS Sequences were grouped into operational taxonomic units (OTUs); we attempted to validate the evolutionary independence of a subset of the inferred OTUs using chloroplast and mitochondrial loci. We explored the patterns of symbiont interactions in these lichens based on ecogeographic distributions and mycobiont taxonomy. We found high levels of undescribed diversity in Trebouxia, broad distributions across distinct ecoregions for many photobiont OTUs and varying levels of mycobiont selectivity and specificity towards the photobiont. Based on these results, we conclude that fungal specificity and selectivity for algal partners play a major role in determining lichen partnerships, potentially superseding ecology, at least at the ecogeographic scale investigated here. To facilitate effective communication and consistency across future studies, we propose a provisional naming system for Trebouxia photobionts and provide Representative Sequences for each OTU circumscribed in this study.
V M Brea - One of the best experts on this subject based on the ideXlab platform.
-
deep learning based multiple object visual tracking on embedded system for iot and mobile edge computing applications
IEEE Internet of Things Journal, 2019Co-Authors: B Blancofilgueira, Daniel Garcialesta, Mauro Fernandezsanjurjo, V M Brea, P LopezAbstract:Compute and memory demands of state-of-the-art deep learning methods are still a shortcoming that must be addressed to make them useful at Internet of Things (IoT) end-nodes. In particular, recent results depict a hopeful prospect for image processing using convolutional neural networks, CNNs, but the gap between software and hardware implementations is already considerable for IoT and mobile edge computing applications due to their high power consumption. This proposal performs low-power and real time deep learning-based multiple object visual tracking implemented on an NVIDIA Jetson TX2 development kit. It includes a camera and wireless connection capability and it is battery powered for mobile and outdoor applications. A collection of Representative Sequences captured with the on-board camera, dETRUSC video dataset, is used to exemplify the performance of the proposed algorithm and to facilitate benchmarking. The results in terms of power consumption and frame rate demonstrate the feasibility of deep learning algorithms on embedded platforms although more effort in the joint algorithm and hardware design of CNNs is needed.
-
deep learning based multiple object visual tracking on embedded system for iot and mobile edge computing applications
arXiv: Computer Vision and Pattern Recognition, 2018Co-Authors: B Blancofilgueira, Daniel Garcialesta, Mauro Fernandezsanjurjo, V M Brea, P LopezAbstract:Compute and memory demands of state-of-the-art deep learning methods are still a shortcoming that must be addressed to make them useful at IoT end-nodes. In particular, recent results depict a hopeful prospect for image processing using Convolutional Neural Netwoks, CNNs, but the gap between software and hardware implementations is already considerable for IoT and mobile edge computing applications due to their high power consumption. This proposal performs low-power and real time deep learning-based multiple object visual tracking implemented on an NVIDIA Jetson TX2 development kit. It includes a camera and wireless connection capability and it is battery powered for mobile and outdoor applications. A collection of Representative Sequences captured with the on-board camera, dETRUSC video dataset, is used to exemplify the performance of the proposed algorithm and to facilitate benchmarking. The results in terms of power consumption and frame rate demonstrate the feasibility of deep learning algorithms on embedded platforms although more effort to joint algorithm and hardware design of CNNs is needed.
