With the fast-developing deep learning models in the field of autonomous driving,the research on the uncertainty estima-tion of deep learning models has also prevailed.Herein,a pyramid Bayesian deep learning method is...With the fast-developing deep learning models in the field of autonomous driving,the research on the uncertainty estima-tion of deep learning models has also prevailed.Herein,a pyramid Bayesian deep learning method is proposed for the model uncertainty evaluation of semantic segmentation.Semantic segmentation is one of the most important perception problems in understanding visual scene,which is critical for autonomous driving.This study to optimize Bayesian SegNet for uncertainty evaluation.This paper first simplifies the network structure of Bayesian SegNet by reducing the number of MC-Dropout layer and then introduces the pyramid pooling module to improve the performance of Bayesian SegNet.mIoU and mPAvPU are used as evaluation matrics to test the proposed method on the public Cityscapes dataset.The experimental results show that the proposed method improves the sampling effect of the Bayesian SegNet,shortens the sampling time,and improves the network performance.展开更多
Geosensing and social sensing as two digitalization mainstreams in big data era are increasingly converging toward an integrated system for the creation of semantically enriched digital Earth.Along with the rapid deve...Geosensing and social sensing as two digitalization mainstreams in big data era are increasingly converging toward an integrated system for the creation of semantically enriched digital Earth.Along with the rapid developments of AI technologies,this convergence has inevitably brought about a number of transformations.On the one hand,value-adding chains from raw data to products and services are becoming value-adding loops composed of four successive stages–Informing,Enabling,Engaging and Empowering(IEEE).Each stage is a dynamic loop for itself.On the other hand,the“human versus technology”relationship is upgraded toward a game-changing“human and technology”collaboration.The information loop is essentially shaped by the omnipresent reciprocity between humans and technologies as equal partners,co-learners and co-creators of new values.The paper gives an analytical review on the mutually changing roles and responsibilities of humans and technologies in the individual stages of the IEEE loop,with the aim to promote a holistic understanding of the state of the art of geospatial information science.Meanwhile,the author elicits a number of challenges facing the interwoven human-technology collaboration.The transformation to a growth mind-set may take time to realize and consolidate.Research works on large-scale semantic data integration are just in the beginning.User experiences of geovisual analytic approaches are far from being systematically studied.Finally,the ethical concerns for the handling of semantically enriched digital Earth cover not only the sensitive issues related to privacy violation,copyright infringement,abuse,etc.but also the questions of how to make technologies as controllable and understandable as possible for humans and how to keep the technological ethos within its constructive sphere of societal influence.展开更多
Metadata are the information about and description of data.In Digital Earth,metadata become variant and heterogeneous with many uncertainties.This paper studies uncertain features in the generation and application of ...Metadata are the information about and description of data.In Digital Earth,metadata become variant and heterogeneous with many uncertainties.This paper studies uncertain features in the generation and application of metadata,and two types of uncertainties(incomplete and imprecise)are described based on semantic quantitative measurement method semantic relationship quantitative measurement based on possibilistic logic and probability statistic(SRQ-PP).Moreover,in the case study,we apply two types of quantitative measurements based on SRQ-PP to describe incomplete(uncertain)knowledge and imprecise(vague)information separately in spatial data service retrieval,which in turn is helpful to identify additional potential data resources and provide a quantitative analysis of the results.展开更多
基金This work was supported by the National Natural Science Foundation of China(U1964203)the National Key R&D Program Project of China(2017YFB0102603).
文摘With the fast-developing deep learning models in the field of autonomous driving,the research on the uncertainty estima-tion of deep learning models has also prevailed.Herein,a pyramid Bayesian deep learning method is proposed for the model uncertainty evaluation of semantic segmentation.Semantic segmentation is one of the most important perception problems in understanding visual scene,which is critical for autonomous driving.This study to optimize Bayesian SegNet for uncertainty evaluation.This paper first simplifies the network structure of Bayesian SegNet by reducing the number of MC-Dropout layer and then introduces the pyramid pooling module to improve the performance of Bayesian SegNet.mIoU and mPAvPU are used as evaluation matrics to test the proposed method on the public Cityscapes dataset.The experimental results show that the proposed method improves the sampling effect of the Bayesian SegNet,shortens the sampling time,and improves the network performance.
基金The figures quoted in this article come from research projects financed by Jiangsu Industrial Technology Research Institute(JITRI),Changshu Fengfan Power Equipment Co.Ltd.,International Graduate School of Science and Engineering(IGSSE)at Technical University of Munich and China Scholarship Council.
文摘Geosensing and social sensing as two digitalization mainstreams in big data era are increasingly converging toward an integrated system for the creation of semantically enriched digital Earth.Along with the rapid developments of AI technologies,this convergence has inevitably brought about a number of transformations.On the one hand,value-adding chains from raw data to products and services are becoming value-adding loops composed of four successive stages–Informing,Enabling,Engaging and Empowering(IEEE).Each stage is a dynamic loop for itself.On the other hand,the“human versus technology”relationship is upgraded toward a game-changing“human and technology”collaboration.The information loop is essentially shaped by the omnipresent reciprocity between humans and technologies as equal partners,co-learners and co-creators of new values.The paper gives an analytical review on the mutually changing roles and responsibilities of humans and technologies in the individual stages of the IEEE loop,with the aim to promote a holistic understanding of the state of the art of geospatial information science.Meanwhile,the author elicits a number of challenges facing the interwoven human-technology collaboration.The transformation to a growth mind-set may take time to realize and consolidate.Research works on large-scale semantic data integration are just in the beginning.User experiences of geovisual analytic approaches are far from being systematically studied.Finally,the ethical concerns for the handling of semantically enriched digital Earth cover not only the sensitive issues related to privacy violation,copyright infringement,abuse,etc.but also the questions of how to make technologies as controllable and understandable as possible for humans and how to keep the technological ethos within its constructive sphere of societal influence.
基金The work in this paper is supported by the National Natural Science Foundation of China under grant no.[61303130]the Natural Science Foundation of Hebei Province under grant no.[F2014203093].
文摘Metadata are the information about and description of data.In Digital Earth,metadata become variant and heterogeneous with many uncertainties.This paper studies uncertain features in the generation and application of metadata,and two types of uncertainties(incomplete and imprecise)are described based on semantic quantitative measurement method semantic relationship quantitative measurement based on possibilistic logic and probability statistic(SRQ-PP).Moreover,in the case study,we apply two types of quantitative measurements based on SRQ-PP to describe incomplete(uncertain)knowledge and imprecise(vague)information separately in spatial data service retrieval,which in turn is helpful to identify additional potential data resources and provide a quantitative analysis of the results.
