Reversible protonic ceramic electrochemical cells(R-PCECs)are ideal,high-effi ciency devices that are environmentally friendly and have a modular design.This paper studies BaFe_(0.6)Zr_(0.1)Y_(0.3)O_(3−δ)(BFZY3)as a ...Reversible protonic ceramic electrochemical cells(R-PCECs)are ideal,high-effi ciency devices that are environmentally friendly and have a modular design.This paper studies BaFe_(0.6)Zr_(0.1)Y_(0.3)O_(3−δ)(BFZY3)as a cobalt-free perovskite oxygen electrode for high-performance R-PCECs where Y ions doping can increase the concentration of oxygen vacancies with a remarkable increase in catalytic performance.The cell with confi guration of Ni-BZCYYb/BZCYYb/BFZY3 demonstrated promising performance in dual modes of fuel cells(FCs)and electrolysis cells(ECs)at 650℃with low polarization resistance of 0.13Ωcm^(2),peak power density of 546.59 mW/cm^(2)in FC mode,and current density of−1.03 A/cm^(2)at 1.3 V in EC mode.The alternative operation between FC and EC modes for up to eight cycles with a total of 80 h suggests that the cell with BFZY3 is exceptionally stable and reversible over the long term.The results indicated that BFZY3 has considerable potential as an air electrode material for R-PCECs,permitting effi cient oxygen reduction and water splitting.展开更多
In this study,we successfully synthesized double perovskite-type oxide NdBa0.5Ca0.5Co1.5Fe0.5O5+δ(NBCCF)using a conventional wet chemical method as the oxygen electrode for reversible solid oxide electrochemical cell...In this study,we successfully synthesized double perovskite-type oxide NdBa0.5Ca0.5Co1.5Fe0.5O5+δ(NBCCF)using a conventional wet chemical method as the oxygen electrode for reversible solid oxide electrochemical cells(RSOCs).The polarization resistance(Rp)of the composite electrode NBCCFGd0.1Ce0.9O2(GDC)is only 0.079Ωcm^2 at 800℃under air.The single cell based on NBCCF-GDC electrode displays a peak power density of 0.941 W/cm^2 in fuel cell mode and a low Rp value of 0.134Ωcm^2.In electrolysis cell mode,the cell displays an outstanding oxygen evolution reaction(OER)activity and shows current density as high as 0.92 A/cm^2 with 50 vol%AH(Absolute Humidity)at 800℃and applied voltage of 1.3 V.Most importantly,the cell exhibits admirable durability of 60 h both in electrolysis mode and fuel cell mode with distinguished reversibility.All these results suggest that NBCCF is a promising candidate electrode for RSOC.展开更多
Objective:To study the effect of lncRNA CTD-2510F5.4 silencing on the proliferation and invasion of huh-7 human hepatocellular carcinoma cells and its possible regulatory relationship with RACGAP1.Methods:The TCGA(The...Objective:To study the effect of lncRNA CTD-2510F5.4 silencing on the proliferation and invasion of huh-7 human hepatocellular carcinoma cells and its possible regulatory relationship with RACGAP1.Methods:The TCGA(The Cancer Genome Atlas)database was used to analyze the expression of lncRNA CTD-2510F5.4 in liver cancer and normal tissues and the correlation with the expression of RACGAP1.CTD-2510F5.4 silencing test was performed on Huh-7 human liver cancer cells,and PCR was used to verify its silencing efficiency.CCK8,Transwell,flow cytometry were used to detect cell proliferation,invasion,and apoptosis,RTPCR,and Western blot.Detect the expression level of RACGAP1.Results:In the TCGA database,lncRNA CTD-2510F5.4 was highly expressed in liver cancer tissues,and it was positively correlated with the expression of RACGAP1(R=0.85).After silencing CTD-2510F5.4,the proliferation and invasion of huh-7 human liver cancer cells were inhibited,apoptosis increased,and the mRNA and protein expression of RACGAP1 were downregulated.Conclusion:lncRNA CTD-2510F5.4 may be the oncogenic lncRNA of liver cancer,and RACGAP1 may be the target of lncRNA CTD-2510F5.4.Silencing CTD-2510F5.4 can inhibit the proliferation and invasion of huh-7 human liver cancer cells,and promote apoptosis,which may be achieved by inhibiting the expression of RACGAP1.展开更多
To promote the electrocatalytic activity and stability of traditional(a_(0.6)Sr_(0.4)Co_(0.2)Fe_(0.8)O_(3-δ)(LSCF)oxygen electrodes in reversible solid oxide cells(RSOCs),conventional physical mixed method was used t...To promote the electrocatalytic activity and stability of traditional(a_(0.6)Sr_(0.4)Co_(0.2)Fe_(0.8)O_(3-δ)(LSCF)oxygen electrodes in reversible solid oxide cells(RSOCs),conventional physical mixed method was used to prepare the Pd-LSCF composite oxygen electrode.The cell with Pd-LSCF|GDC|YSZ|Ni-YSZ configuration shows perfect electrochemical performance in both solid oxide fuel cell(SOFC)mode and solid oxide electrolysis cell(SOEC)mode.In the SOFC mode,the cell achieves a power density of 1.73 W/cm^(2)at800℃higher than that of the LSCF oxygen electrode with 1.38 W/cm^(2).In the SOEC mode,the current density at 1.5 V is 1.67 A/cm^(2)at 800℃under 50 vol%steam concentration.Moreover,the reversibility and stability of the RSOCs were tested during 192 h long-term reversible operation.The degradation rate of the cell is only 2.2%/100 h and 2.5%/100 h in the SOEC and the SOFC modes,respectively.These results confirm that compositing Pd with the LSCF oxygen electrode can considerably boost the electrochemical performance of LSCF electrode in RSOCs field.展开更多
As one of the most fundamental topics in reinforcement learning(RL),sample efficiency is essential to the deployment of deep RL algorithms.Unlike most existing exploration methods that sample an action from different ...As one of the most fundamental topics in reinforcement learning(RL),sample efficiency is essential to the deployment of deep RL algorithms.Unlike most existing exploration methods that sample an action from different types of posterior distributions,we focus on the policy sampling process and propose an efficient selective sampling approach to improve sample efficiency by modeling the internal hierarchy of the environment.Specifically,we first employ clustering methods in the policy sampling process to generate an action candidate set.Then we introduce a clustering buffer for modeling the internal hierarchy,which consists of on-policy data,off-policy data,and expert data to evaluate actions from the clusters in the action candidate set in the exploration stage.In this way,our approach is able to take advantage of the supervision information in the expert demonstration data.Experiments on six different continuous locomotion environments demonstrate superior reinforcement learning performance and faster convergence of selective sampling.In particular,on the LGSVL task,our method can reduce the number of convergence steps by 46.7%and the convergence time by 28.5%.Furthermore,our code is open-source for reproducibility.The code is available at https://github.com/Shihwin/SelectiveSampling.展开更多
Corrosion at the taper/trunnion interface of total hip replacement(THR)often results in severe complications.However,the underlying mechanisms of biotribocorrosion at the taper/trunnion interface during the long-term ...Corrosion at the taper/trunnion interface of total hip replacement(THR)often results in severe complications.However,the underlying mechanisms of biotribocorrosion at the taper/trunnion interface during the long-term walking gait cycles remain to be fully understood.In this study,a hip joint simulator was therefore instrumented with an electrochemical cell for in-situ monitoring of the tribocorrosion evolution in a metal-on-polyethylene(MoP)THR during a typical long-term walking gait.In addition,the biotribocorrosion mechanism was investigated via surface and chemical characterizations.The experimental results confirmed that the taper/trunnion interface dominated the contemporary MoP hip joint corrosion.Three cyclic variations in the open circuit potential(OCP)were observed throughout the long-term electrochemical measurements,attributed to the formation and disruption of the adsorbed protein layer.The corrosion exhibited an initial increase at each period,peaking at approximately 0.125 million cycles,followed by a subsequent gradual reduction.Surface and chemical analyses revealed the formation of a tribochemical reaction layer(tribolayer)on the worn surface of the taper/trunnion interface.The surface/chemical characterizations and the electrochemical measurements indicated that the adhesion force of the adsorbed protein layer was weaker than that of the tribolayer.In contrast,the opposite was true for the corrosion resistance.Based on the observations from this study,the tribocorrosion mechanism of the taper/trunnion interface under the long-term walking gait cycles is deduced.展开更多
Urban development continues to reduce the amount of available ground space.The development of underground space is thus gath-ering increasing attention to alleviate ground congestion.However,there is currently a lack ...Urban development continues to reduce the amount of available ground space.The development of underground space is thus gath-ering increasing attention to alleviate ground congestion.However,there is currently a lack of a three-dimensional(3D)evaluation method to systematically evaluate the geological conditions of underground space and possible geological disaster risks caused by rock and soil masses.This paper presents an engineering geological suitability assessment framework based on 3D geological modeling and an analytic hierarchy process(AHP)-cloud model.As the basis for 3D evaluation,a 3D structural model of the study area is established based on the drilling data and geological profiles.Then the structural model is partitioned to obtain interpolation grids,and the ordinary Kriging interpolation method is applied to attribute interpolation.All the attributes are exported from the geological model,and the rock and soil masses are divided into four categories according to their engineering properties,namely soft soil,sandy soil,cohesive soil,and rock,upon which a targeted hierarchy structure is established based on the attributes that impact the suitability.This paper intro-duces the cloud model to characterize the uncertainty of these evaluation indexes,which synthesizes an AHP method,thus it is referred to as the AHP-cloud model.This new model is used to evaluate the geological suitability of underground space in the Sanlong Bay district,Foshan City,Guangdong,China.In addition,we also determine the excavation difficulty at different depths according to the lithology and weathering degree of the study area.The limitations and future directions of the proposed method are discussed,including the influ-encing factors and weight determination.展开更多
Hydrogen energy,the cleanest fuel,presents extensive applications in renewable energy technologies such as fuel cells.However,the transition process from carbon-based(fossil fuel)energy to desired hydrogen energy is u...Hydrogen energy,the cleanest fuel,presents extensive applications in renewable energy technologies such as fuel cells.However,the transition process from carbon-based(fossil fuel)energy to desired hydrogen energy is usually hindered by inevitable scientific,technological,and economic obstacles,which mainly involves complex hydrocarbon reforming reactions.Hence,this paper provides a systematic and comprehensive analysis focusing on the hydrocarbon reforming mechanism.Accordingly,recent related studies are summarized to clarify the intrinsic difference among the reforming mechanism.Aiming to objectively assess the activated catalyst and deactivation mechanism,the rate-determining steps of reforming process have been emphasized,summarized,and analyzed.Specifically,the effect of metals and supports on individual reaction processes is discussed followed by the metalsupport interaction.Current tendency and research map could be established to promote the technology development and expansion of hydrocarbon reforming field.This review could be considered as the guideline for academics and industry designing appropriate catalysts.展开更多
Sepsis treatment is a highly challenging effort to reduce mortality in hospital intensive care units since the treatment response may vary for each patient.Tailored treatment recommendations are desired to assist doct...Sepsis treatment is a highly challenging effort to reduce mortality in hospital intensive care units since the treatment response may vary for each patient.Tailored treatment recommendations are desired to assist doctors in making decisions efficiently and accurately.In this work,we apply a self-supervised method based on reinforcement learning(RL)for treatment recommendation on individuals.An uncertainty evaluation method is proposed to separate patient samples into two domains according to their responses to treatments and the state value of the chosen policy.Examples of two domains are then reconstructed with an auxiliary transfer learning task.A distillation method of privilege learning is tied to a variational auto-encoder framework for the transfer learning task between the low-and high-quality domains.Combined with the self-supervised way for better state and action representations,we propose a deep RL method called high-risk uncertainty(HRU)control to provide flexibility on the trade-off between the effectiveness and accuracy of ambiguous samples and to reduce the expected mortality.Experiments on the large-scale publicly available real-world dataset MIMIC-Ⅲdemonstrate that our model reduces the estimated mortality rate by up to 2.3%in total,and that the estimated mortality rate in the majority of cases is reduced to 9.5%.展开更多
Crowd counting has been applied to a variety of applications such as video surveillance,traffic monitoring,assembly control,and other public safety applications.Context information,such as perspective distortion and b...Crowd counting has been applied to a variety of applications such as video surveillance,traffic monitoring,assembly control,and other public safety applications.Context information,such as perspective distortion and background interference,is a crucial factor in achieving high performance for crowd counting.While traditional methods focus merely on solving one specific factor,we aggregate sufficient context information into the crowd counting network to tackle these problems simultaneously in this study.We build a fully convolutional network with two tasks,i.e.,main density map estimation and auxiliary semantic segmentation.The main task is to extract the multi-scale and spatial context information to learn the density map.The auxiliary semantic segmentation task gives a comprehensive view of the background and foreground information,and the extracted information is finally incorporated into the main task by late fusion.We demonstrate that our network has better accuracy of estimation and higher robustness on three challenging datasets compared with state-of-the-art methods.展开更多
基金support from the National Key Research&Development Project(2022YFB4002201)National Natural Science Foundation of China(Nos.52172199,52072135,52002121)+1 种基金Hubei Province(2023BAB115)Jiangsu Province(BZ2022027).
文摘Reversible protonic ceramic electrochemical cells(R-PCECs)are ideal,high-effi ciency devices that are environmentally friendly and have a modular design.This paper studies BaFe_(0.6)Zr_(0.1)Y_(0.3)O_(3−δ)(BFZY3)as a cobalt-free perovskite oxygen electrode for high-performance R-PCECs where Y ions doping can increase the concentration of oxygen vacancies with a remarkable increase in catalytic performance.The cell with confi guration of Ni-BZCYYb/BZCYYb/BFZY3 demonstrated promising performance in dual modes of fuel cells(FCs)and electrolysis cells(ECs)at 650℃with low polarization resistance of 0.13Ωcm^(2),peak power density of 546.59 mW/cm^(2)in FC mode,and current density of−1.03 A/cm^(2)at 1.3 V in EC mode.The alternative operation between FC and EC modes for up to eight cycles with a total of 80 h suggests that the cell with BFZY3 is exceptionally stable and reversible over the long term.The results indicated that BFZY3 has considerable potential as an air electrode material for R-PCECs,permitting effi cient oxygen reduction and water splitting.
基金financial support from National Key Research&Development Project(2016YFE0126900)the National Natural Science Foundation of China(51672095)+2 种基金Hubei Province(2018AAA057)the EPSRC Capital for Great Technologies Grant EP/L017008/1the China Scholarship Council for funding(201806160178)。
文摘In this study,we successfully synthesized double perovskite-type oxide NdBa0.5Ca0.5Co1.5Fe0.5O5+δ(NBCCF)using a conventional wet chemical method as the oxygen electrode for reversible solid oxide electrochemical cells(RSOCs).The polarization resistance(Rp)of the composite electrode NBCCFGd0.1Ce0.9O2(GDC)is only 0.079Ωcm^2 at 800℃under air.The single cell based on NBCCF-GDC electrode displays a peak power density of 0.941 W/cm^2 in fuel cell mode and a low Rp value of 0.134Ωcm^2.In electrolysis cell mode,the cell displays an outstanding oxygen evolution reaction(OER)activity and shows current density as high as 0.92 A/cm^2 with 50 vol%AH(Absolute Humidity)at 800℃and applied voltage of 1.3 V.Most importantly,the cell exhibits admirable durability of 60 h both in electrolysis mode and fuel cell mode with distinguished reversibility.All these results suggest that NBCCF is a promising candidate electrode for RSOC.
基金Baise City Scientific Research and Technology Development Plan(No.20195405)。
文摘Objective:To study the effect of lncRNA CTD-2510F5.4 silencing on the proliferation and invasion of huh-7 human hepatocellular carcinoma cells and its possible regulatory relationship with RACGAP1.Methods:The TCGA(The Cancer Genome Atlas)database was used to analyze the expression of lncRNA CTD-2510F5.4 in liver cancer and normal tissues and the correlation with the expression of RACGAP1.CTD-2510F5.4 silencing test was performed on Huh-7 human liver cancer cells,and PCR was used to verify its silencing efficiency.CCK8,Transwell,flow cytometry were used to detect cell proliferation,invasion,and apoptosis,RTPCR,and Western blot.Detect the expression level of RACGAP1.Results:In the TCGA database,lncRNA CTD-2510F5.4 was highly expressed in liver cancer tissues,and it was positively correlated with the expression of RACGAP1(R=0.85).After silencing CTD-2510F5.4,the proliferation and invasion of huh-7 human liver cancer cells were inhibited,apoptosis increased,and the mRNA and protein expression of RACGAP1 were downregulated.Conclusion:lncRNA CTD-2510F5.4 may be the oncogenic lncRNA of liver cancer,and RACGAP1 may be the target of lncRNA CTD-2510F5.4.Silencing CTD-2510F5.4 can inhibit the proliferation and invasion of huh-7 human liver cancer cells,and promote apoptosis,which may be achieved by inhibiting the expression of RACGAP1.
基金Project supported by the National Key Research&Development Project(2020YFB1506304)the National Natural Science Foundation of China(52172199,52072135,52002121)。
文摘To promote the electrocatalytic activity and stability of traditional(a_(0.6)Sr_(0.4)Co_(0.2)Fe_(0.8)O_(3-δ)(LSCF)oxygen electrodes in reversible solid oxide cells(RSOCs),conventional physical mixed method was used to prepare the Pd-LSCF composite oxygen electrode.The cell with Pd-LSCF|GDC|YSZ|Ni-YSZ configuration shows perfect electrochemical performance in both solid oxide fuel cell(SOFC)mode and solid oxide electrolysis cell(SOEC)mode.In the SOFC mode,the cell achieves a power density of 1.73 W/cm^(2)at800℃higher than that of the LSCF oxygen electrode with 1.38 W/cm^(2).In the SOEC mode,the current density at 1.5 V is 1.67 A/cm^(2)at 800℃under 50 vol%steam concentration.Moreover,the reversibility and stability of the RSOCs were tested during 192 h long-term reversible operation.The degradation rate of the cell is only 2.2%/100 h and 2.5%/100 h in the SOEC and the SOFC modes,respectively.These results confirm that compositing Pd with the LSCF oxygen electrode can considerably boost the electrochemical performance of LSCF electrode in RSOCs field.
基金supported by the National Natural Science Foundation of China (No.62176059)the Shanghai Municipal Science and Technology Major Project (No.2018SHZDZX01)Zhangjiang Lab,and the Shanghai Center for Brain Science and Brain-inspired Technology。
文摘As one of the most fundamental topics in reinforcement learning(RL),sample efficiency is essential to the deployment of deep RL algorithms.Unlike most existing exploration methods that sample an action from different types of posterior distributions,we focus on the policy sampling process and propose an efficient selective sampling approach to improve sample efficiency by modeling the internal hierarchy of the environment.Specifically,we first employ clustering methods in the policy sampling process to generate an action candidate set.Then we introduce a clustering buffer for modeling the internal hierarchy,which consists of on-policy data,off-policy data,and expert data to evaluate actions from the clusters in the action candidate set in the exploration stage.In this way,our approach is able to take advantage of the supervision information in the expert demonstration data.Experiments on six different continuous locomotion environments demonstrate superior reinforcement learning performance and faster convergence of selective sampling.In particular,on the LGSVL task,our method can reduce the number of convergence steps by 46.7%and the convergence time by 28.5%.Furthermore,our code is open-source for reproducibility.The code is available at https://github.com/Shihwin/SelectiveSampling.
基金supported by the National Natural Science Foundation of China(52035012)the Science and Technology Planning Project of Sichuan Province(2020YJ0032)the 111 Project(B20008).
文摘Corrosion at the taper/trunnion interface of total hip replacement(THR)often results in severe complications.However,the underlying mechanisms of biotribocorrosion at the taper/trunnion interface during the long-term walking gait cycles remain to be fully understood.In this study,a hip joint simulator was therefore instrumented with an electrochemical cell for in-situ monitoring of the tribocorrosion evolution in a metal-on-polyethylene(MoP)THR during a typical long-term walking gait.In addition,the biotribocorrosion mechanism was investigated via surface and chemical characterizations.The experimental results confirmed that the taper/trunnion interface dominated the contemporary MoP hip joint corrosion.Three cyclic variations in the open circuit potential(OCP)were observed throughout the long-term electrochemical measurements,attributed to the formation and disruption of the adsorbed protein layer.The corrosion exhibited an initial increase at each period,peaking at approximately 0.125 million cycles,followed by a subsequent gradual reduction.Surface and chemical analyses revealed the formation of a tribochemical reaction layer(tribolayer)on the worn surface of the taper/trunnion interface.The surface/chemical characterizations and the electrochemical measurements indicated that the adhesion force of the adsorbed protein layer was weaker than that of the tribolayer.In contrast,the opposite was true for the corrosion resistance.Based on the observations from this study,the tribocorrosion mechanism of the taper/trunnion interface under the long-term walking gait cycles is deduced.
基金the Foshan Urban Geological Survey Pilot:Urban geological survey of the Boot Area of Sanlong Bay High-end Innovation Cluster Areas(Grant No.440600-202004-211001-0011).
文摘Urban development continues to reduce the amount of available ground space.The development of underground space is thus gath-ering increasing attention to alleviate ground congestion.However,there is currently a lack of a three-dimensional(3D)evaluation method to systematically evaluate the geological conditions of underground space and possible geological disaster risks caused by rock and soil masses.This paper presents an engineering geological suitability assessment framework based on 3D geological modeling and an analytic hierarchy process(AHP)-cloud model.As the basis for 3D evaluation,a 3D structural model of the study area is established based on the drilling data and geological profiles.Then the structural model is partitioned to obtain interpolation grids,and the ordinary Kriging interpolation method is applied to attribute interpolation.All the attributes are exported from the geological model,and the rock and soil masses are divided into four categories according to their engineering properties,namely soft soil,sandy soil,cohesive soil,and rock,upon which a targeted hierarchy structure is established based on the attributes that impact the suitability.This paper intro-duces the cloud model to characterize the uncertainty of these evaluation indexes,which synthesizes an AHP method,thus it is referred to as the AHP-cloud model.This new model is used to evaluate the geological suitability of underground space in the Sanlong Bay district,Foshan City,Guangdong,China.In addition,we also determine the excavation difficulty at different depths according to the lithology and weathering degree of the study area.The limitations and future directions of the proposed method are discussed,including the influ-encing factors and weight determination.
基金This work was financially supported by National Key Research&Development Project of China[2022YFB4002203]National Natural Science Foundation of China[52072135,22005227].
文摘Hydrogen energy,the cleanest fuel,presents extensive applications in renewable energy technologies such as fuel cells.However,the transition process from carbon-based(fossil fuel)energy to desired hydrogen energy is usually hindered by inevitable scientific,technological,and economic obstacles,which mainly involves complex hydrocarbon reforming reactions.Hence,this paper provides a systematic and comprehensive analysis focusing on the hydrocarbon reforming mechanism.Accordingly,recent related studies are summarized to clarify the intrinsic difference among the reforming mechanism.Aiming to objectively assess the activated catalyst and deactivation mechanism,the rate-determining steps of reforming process have been emphasized,summarized,and analyzed.Specifically,the effect of metals and supports on individual reaction processes is discussed followed by the metalsupport interaction.Current tendency and research map could be established to promote the technology development and expansion of hydrocarbon reforming field.This review could be considered as the guideline for academics and industry designing appropriate catalysts.
基金the National Natural Science Foundation of China(No.61702186)。
文摘Sepsis treatment is a highly challenging effort to reduce mortality in hospital intensive care units since the treatment response may vary for each patient.Tailored treatment recommendations are desired to assist doctors in making decisions efficiently and accurately.In this work,we apply a self-supervised method based on reinforcement learning(RL)for treatment recommendation on individuals.An uncertainty evaluation method is proposed to separate patient samples into two domains according to their responses to treatments and the state value of the chosen policy.Examples of two domains are then reconstructed with an auxiliary transfer learning task.A distillation method of privilege learning is tied to a variational auto-encoder framework for the transfer learning task between the low-and high-quality domains.Combined with the self-supervised way for better state and action representations,we propose a deep RL method called high-risk uncertainty(HRU)control to provide flexibility on the trade-off between the effectiveness and accuracy of ambiguous samples and to reduce the expected mortality.Experiments on the large-scale publicly available real-world dataset MIMIC-Ⅲdemonstrate that our model reduces the estimated mortality rate by up to 2.3%in total,and that the estimated mortality rate in the majority of cases is reduced to 9.5%.
基金the National Natural Science Foundation of China(Nos.61702186,61672236,and 61602459)。
文摘Crowd counting has been applied to a variety of applications such as video surveillance,traffic monitoring,assembly control,and other public safety applications.Context information,such as perspective distortion and background interference,is a crucial factor in achieving high performance for crowd counting.While traditional methods focus merely on solving one specific factor,we aggregate sufficient context information into the crowd counting network to tackle these problems simultaneously in this study.We build a fully convolutional network with two tasks,i.e.,main density map estimation and auxiliary semantic segmentation.The main task is to extract the multi-scale and spatial context information to learn the density map.The auxiliary semantic segmentation task gives a comprehensive view of the background and foreground information,and the extracted information is finally incorporated into the main task by late fusion.We demonstrate that our network has better accuracy of estimation and higher robustness on three challenging datasets compared with state-of-the-art methods.