Accurate and continuous identification of individual cattle is crucial to precision farming in recent years.It is also the prerequisite to monitor the individual feed intake and feeding time of beef cattle at medium t...Accurate and continuous identification of individual cattle is crucial to precision farming in recent years.It is also the prerequisite to monitor the individual feed intake and feeding time of beef cattle at medium to long distances over different cameras.However,beef cattle can tend to frequently move and change their feeding position during feeding.Furthermore,the great variations in their head direction and complex environments(light,occlusion,and background)can also lead to some difficulties in the recognition,particularly for the bio-similarities among individual cattle.Among them,AlignedReID++model is characterized by both global and local information for image matching.In particular,the dynamically matching local information(DMLI)algorithm has been introduced into the local branch to automatically align the horizontal local information.In this research,the AlignedReID++model was utilized and improved to achieve the better performance in cattle re-identification(ReID).Initially,triplet attention(TA)modules were integrated into the BottleNecks of ResNet50 Backbone.The feature extraction was then enhanced through cross-dimensional interactions with the minimal computational overhead.Since the TA modules in AlignedReID++baseline model increased the model size and floating point operations(FLOPs)by 0.005 M and 0.05 G,the rank-1 accuracy and mean average precision(mAP)were improved by 1.0 percentage points and 2.94 percentage points,respectively.Specifically,the rank-1 accuracies were outperformed by 0.86 percentage points and 0.12 percentage points,respectively,compared with the convolution block attention module(CBAM)and efficient channel attention(ECA)modules,although 0.94 percentage points were lower than that of squeeze-and-excitation(SE)modules.The mAP metric values were exceeded by 0.22,0.86 and 0.12 percentage points,respectively,compared with the SE,CBAM,and ECA modules.Additionally,the Cross-Entropy Loss function was replaced with the CosFace Loss function in the global branch of baseline model.CosFace Loss and Hard Triplet Loss were jointly employed to train the baseline model for the better identification on the similar individuals.AlignedReID++with CosFace Loss was outperformed the baseline model by 0.24 and 0.92 percentage points in the rank-1 accuracy and mAP,respectively,whereas,AlignedReID++with ArcFace Loss was exceeded by 0.36 and 0.56 percentage points,respectively.The improved model with the TA modules and CosFace Loss was achieved in a rank-1 accuracy of 94.42%,rank-5 accuracy of 98.78%,rank-10 accuracy of 99.34%,mAP of 63.90%,FLOPs of 5.45 G,frames per second(FPS)of 5.64,and model size of 23.78 M.The rank-1 accuracies were exceeded by 1.84,4.72,0.76 and 5.36 percentage points,respectively,compared with the baseline model,part-based convolutional baseline(PCB),multiple granularity network(MGN),and relation-aware global attention(RGA),while the mAP metrics were surpassed 6.42,5.86,4.30 and 7.38 percentage points,respectively.Meanwhile,the rank-1 accuracy was 0.98 percentage points lower than TransReID,but the mAP metric was exceeded by 3.90 percentage points.Moreover,the FLOPs of improved model were only 0.05 G larger than that of baseline model,while smaller than those of PCB,MGN,RGA,and TransReID by 0.68,6.51,25.4,and 16.55 G,respectively.The model size of improved model was 23.78 M,which was smaller than those of the baseline model,PCB,MGN,RGA,and TransReID by 0.03,2.33,45.06,14.53 and 62.85 M,respectively.The inference speed of improved model on a CPU was lower than those of PCB,MGN,and baseline model,but higher than TransReID and RGA.The t-SNE feature embedding visualization demonstrated that the global and local features were achieve in the better intra-class compactness and inter-class variability.Therefore,the improved model can be expected to effectively re-identify the beef cattle in natural environments of breeding farm,in order to monitor the individual feed intake and feeding time.展开更多
The advent of gene editing represents one of the most transformative breakthroughs in life science,making genome manipulation more accessible than ever before.While traditional CRISPR/Cas-based gene editing,which invo...The advent of gene editing represents one of the most transformative breakthroughs in life science,making genome manipulation more accessible than ever before.While traditional CRISPR/Cas-based gene editing,which involves double-strand DNA breaks(DSBs),excels at gene disruption,it is less effective for accurate gene modification.The limitation arises because DSBs are primarily repaired via non-homologous end joining(NHEJ),which tends to introduce indels at the break site.While homology directed repair(HDR)can achieve precise editing when a donor DNA template is provided,the reliance on DSBs often results in unintended genome damage.HDR is restricted to specific cell cycle phases,limiting its application.Currently,gene editing has evolved to unprecedented levels of precision without relying on DSB and HDR.The development of innovative systems,such as base editing,prime editing,and CRISPR-associated transposases(CASTs),now allow for precise editing ranging from single nucleotides to large DNA fragments.Base editors(BEs)enable the direct conversion of one nucleotide to another,and prime editors(PEs)further expand gene editing capabilities by allowing for the insertion,deletion,or alteration of small DNA fragments.The CAST system,a recent innovation,allows for the precise insertion of large DNA fragments at specific genomic locations.In recent years,the optimization of these precise gene editing tools has led to significant improvements in editing efficiency,specificity,and versatility,with advancements such as the creation of base editors for nucleotide transversions,enhanced prime editing systems for more efficient and precise modifications,and refined CAST systems for targeted large DNA insertions,expanding the range of applications for these tools.Concurrently,these advances are complemented by significant improvements in in vivo delivery methods,which have paved the way for therapeutic application of precise gene editing tools.Effective delivery systems are critical for the success of gene therapies,and recent developments in both viral and non-viral vectors have improved the efficiency and safety of gene editing.For instance,adeno-associated viruses(AAVs)are widely used due to their high transfection efficiency and low immunogenicity,though challenges such as limited cargo capacity and potential for immune responses remain.Non-viral delivery systems,including lipid nanoparticles(LNPs),offer an alternative with lower immunogenicity and higher payload capacity,although their transfection efficiency can be lower.The therapeutic potential of these precise gene editing technologies is vast,particularly in treating genetic disorders.Preclinical studies have demonstrated the effectiveness of base editing in correcting genetic mutations responsible for diseases such as cardiomyopathy,liver disease,and hereditary hearing loss.These technologies promise to treat symptoms and potentially cure the underlying genetic causes of these conditions.Meanwhile,challenges remain,such as optimizing the safety and specificity of gene editing tools,improving delivery systems,and overcoming off-target effects,all of which are critical for their successful application in clinical settings.In summary,the continuous evolution of precise gene editing technologies,combined with advancements in delivery systems,is driving the field toward new therapeutic applications that can potentially transform the treatment of genetic disorders by targeting their root causes.展开更多
Silicon(Si)diffraction microlens arrays are usually used to integrating with infrared focal plane arrays(IRFPAs)to improve their performance.The errors of lithography are unavoidable in the process of the Si diffrac-t...Silicon(Si)diffraction microlens arrays are usually used to integrating with infrared focal plane arrays(IRFPAs)to improve their performance.The errors of lithography are unavoidable in the process of the Si diffrac-tion microlens arrays preparation in the conventional engraving method.It has a serious impact on its performance and subsequent applications.In response to the problem of errors of Si diffraction microlens arrays in the conven-tional method,a novel self-alignment method for high precision Si diffraction microlens arrays preparation is pro-posed.The accuracy of the Si diffractive microlens arrays preparation is determined by the accuracy of the first li-thography mask in the novel self-alignment method.In the subsequent etching,the etched area will be protected by the mask layer and the sacrifice layer or the protective layer.The unprotection area is carved to effectively block the non-etching areas,accurately etch the etching area required,and solve the problem of errors.The high precision Si diffraction microlens arrays are obtained by the novel self-alignment method and the diffraction effi-ciency could reach 92.6%.After integrating with IRFPAs,the average blackbody responsity increased by 8.3%,and the average blackbody detectivity increased by 10.3%.It indicates that the Si diffraction microlens arrays can improve the filling factor and reduce crosstalk of IRFPAs through convergence,thereby improving the perfor-mance of the IRFPAs.The results are of great reference significance for improving their performance through opti-mizing the preparation level of micro nano devices.展开更多
The main principle and mathematical model of GOCE kinematic orbit adjustment for Earth gravity field model （EGM） validation and accelerometer calibration are presented. Based on 60 days GOCE kinematic orbits with 1-...The main principle and mathematical model of GOCE kinematic orbit adjustment for Earth gravity field model （EGM） validation and accelerometer calibration are presented. Based on 60 days GOCE kinematic orbits with 1-2 cm accuracy and accelerometer data from 2009-11-02 to 2009-12-31, the RMS-of-fit （ROF） of them using EGM2008, EIGEN-SC, ITG- GRACE2010S and GOCO01S up to 120, 150 and 180 degree and order （d/o） are evaluated and compared. The scale factors and biases of GOCE accelerometer data are calibrated and the energy balance method （EBM） is performed to test the accuracy of accelerometer calibration. The results show that GOCE orbits are also sensitive to EGM from 120 to 150 d/o. The ROFs of EGMs with 150 and 180 d/o are obviously better than those of EGMs with 120 d/o. The ROFs of GOCO01S and ITG-GRACE2010S are almost the same up to 120 and 150 d/o, which are about 3.3 cm and 1.8 cm, respectively. They are far better than those of EGM2008 and EIGEN-SC with the same d/o. The ROF of GOCO01S with 180 d/o is about 1.6 em, which is the best one among those EGMs. The accelerometer calibration accuracies （ACAs） of ITG-GRACE2010S and GOCO01S are obviously higher that those of EGM2008 and EIGEN-SC. The ACA of GOCO01S with 180 d/o is far higher than that of EGMs with 120 d/o, and a little higher than that of ITG-GRACE2010S with 150 d/o. I t is suggested that the newest released EGM such as GOCO01S or GOCO02S till at least 150 d/o should be chosen in GOCE precise orbit determination （POD） and accelerometer calibration.展开更多
基金National Key Research and Development Program(2023YFD1301801)National Natural Science Foundation of China(32272931)+1 种基金Shaanxi Province Agricultural Key Core Technology Project(2024NYGG005)Shaanxi Province Key R&D Program(2024NC-ZDCYL-05-12)。
文摘Accurate and continuous identification of individual cattle is crucial to precision farming in recent years.It is also the prerequisite to monitor the individual feed intake and feeding time of beef cattle at medium to long distances over different cameras.However,beef cattle can tend to frequently move and change their feeding position during feeding.Furthermore,the great variations in their head direction and complex environments(light,occlusion,and background)can also lead to some difficulties in the recognition,particularly for the bio-similarities among individual cattle.Among them,AlignedReID++model is characterized by both global and local information for image matching.In particular,the dynamically matching local information(DMLI)algorithm has been introduced into the local branch to automatically align the horizontal local information.In this research,the AlignedReID++model was utilized and improved to achieve the better performance in cattle re-identification(ReID).Initially,triplet attention(TA)modules were integrated into the BottleNecks of ResNet50 Backbone.The feature extraction was then enhanced through cross-dimensional interactions with the minimal computational overhead.Since the TA modules in AlignedReID++baseline model increased the model size and floating point operations(FLOPs)by 0.005 M and 0.05 G,the rank-1 accuracy and mean average precision(mAP)were improved by 1.0 percentage points and 2.94 percentage points,respectively.Specifically,the rank-1 accuracies were outperformed by 0.86 percentage points and 0.12 percentage points,respectively,compared with the convolution block attention module(CBAM)and efficient channel attention(ECA)modules,although 0.94 percentage points were lower than that of squeeze-and-excitation(SE)modules.The mAP metric values were exceeded by 0.22,0.86 and 0.12 percentage points,respectively,compared with the SE,CBAM,and ECA modules.Additionally,the Cross-Entropy Loss function was replaced with the CosFace Loss function in the global branch of baseline model.CosFace Loss and Hard Triplet Loss were jointly employed to train the baseline model for the better identification on the similar individuals.AlignedReID++with CosFace Loss was outperformed the baseline model by 0.24 and 0.92 percentage points in the rank-1 accuracy and mAP,respectively,whereas,AlignedReID++with ArcFace Loss was exceeded by 0.36 and 0.56 percentage points,respectively.The improved model with the TA modules and CosFace Loss was achieved in a rank-1 accuracy of 94.42%,rank-5 accuracy of 98.78%,rank-10 accuracy of 99.34%,mAP of 63.90%,FLOPs of 5.45 G,frames per second(FPS)of 5.64,and model size of 23.78 M.The rank-1 accuracies were exceeded by 1.84,4.72,0.76 and 5.36 percentage points,respectively,compared with the baseline model,part-based convolutional baseline(PCB),multiple granularity network(MGN),and relation-aware global attention(RGA),while the mAP metrics were surpassed 6.42,5.86,4.30 and 7.38 percentage points,respectively.Meanwhile,the rank-1 accuracy was 0.98 percentage points lower than TransReID,but the mAP metric was exceeded by 3.90 percentage points.Moreover,the FLOPs of improved model were only 0.05 G larger than that of baseline model,while smaller than those of PCB,MGN,RGA,and TransReID by 0.68,6.51,25.4,and 16.55 G,respectively.The model size of improved model was 23.78 M,which was smaller than those of the baseline model,PCB,MGN,RGA,and TransReID by 0.03,2.33,45.06,14.53 and 62.85 M,respectively.The inference speed of improved model on a CPU was lower than those of PCB,MGN,and baseline model,but higher than TransReID and RGA.The t-SNE feature embedding visualization demonstrated that the global and local features were achieve in the better intra-class compactness and inter-class variability.Therefore,the improved model can be expected to effectively re-identify the beef cattle in natural environments of breeding farm,in order to monitor the individual feed intake and feeding time.
文摘The advent of gene editing represents one of the most transformative breakthroughs in life science,making genome manipulation more accessible than ever before.While traditional CRISPR/Cas-based gene editing,which involves double-strand DNA breaks(DSBs),excels at gene disruption,it is less effective for accurate gene modification.The limitation arises because DSBs are primarily repaired via non-homologous end joining(NHEJ),which tends to introduce indels at the break site.While homology directed repair(HDR)can achieve precise editing when a donor DNA template is provided,the reliance on DSBs often results in unintended genome damage.HDR is restricted to specific cell cycle phases,limiting its application.Currently,gene editing has evolved to unprecedented levels of precision without relying on DSB and HDR.The development of innovative systems,such as base editing,prime editing,and CRISPR-associated transposases(CASTs),now allow for precise editing ranging from single nucleotides to large DNA fragments.Base editors(BEs)enable the direct conversion of one nucleotide to another,and prime editors(PEs)further expand gene editing capabilities by allowing for the insertion,deletion,or alteration of small DNA fragments.The CAST system,a recent innovation,allows for the precise insertion of large DNA fragments at specific genomic locations.In recent years,the optimization of these precise gene editing tools has led to significant improvements in editing efficiency,specificity,and versatility,with advancements such as the creation of base editors for nucleotide transversions,enhanced prime editing systems for more efficient and precise modifications,and refined CAST systems for targeted large DNA insertions,expanding the range of applications for these tools.Concurrently,these advances are complemented by significant improvements in in vivo delivery methods,which have paved the way for therapeutic application of precise gene editing tools.Effective delivery systems are critical for the success of gene therapies,and recent developments in both viral and non-viral vectors have improved the efficiency and safety of gene editing.For instance,adeno-associated viruses(AAVs)are widely used due to their high transfection efficiency and low immunogenicity,though challenges such as limited cargo capacity and potential for immune responses remain.Non-viral delivery systems,including lipid nanoparticles(LNPs),offer an alternative with lower immunogenicity and higher payload capacity,although their transfection efficiency can be lower.The therapeutic potential of these precise gene editing technologies is vast,particularly in treating genetic disorders.Preclinical studies have demonstrated the effectiveness of base editing in correcting genetic mutations responsible for diseases such as cardiomyopathy,liver disease,and hereditary hearing loss.These technologies promise to treat symptoms and potentially cure the underlying genetic causes of these conditions.Meanwhile,challenges remain,such as optimizing the safety and specificity of gene editing tools,improving delivery systems,and overcoming off-target effects,all of which are critical for their successful application in clinical settings.In summary,the continuous evolution of precise gene editing technologies,combined with advancements in delivery systems,is driving the field toward new therapeutic applications that can potentially transform the treatment of genetic disorders by targeting their root causes.
基金Supported by the National Natural Science Foundation of China(NSFC 62105100)the National Key research and development program in the 14th five year plan(2021YFA1200700)。
文摘Silicon(Si)diffraction microlens arrays are usually used to integrating with infrared focal plane arrays(IRFPAs)to improve their performance.The errors of lithography are unavoidable in the process of the Si diffrac-tion microlens arrays preparation in the conventional engraving method.It has a serious impact on its performance and subsequent applications.In response to the problem of errors of Si diffraction microlens arrays in the conven-tional method,a novel self-alignment method for high precision Si diffraction microlens arrays preparation is pro-posed.The accuracy of the Si diffractive microlens arrays preparation is determined by the accuracy of the first li-thography mask in the novel self-alignment method.In the subsequent etching,the etched area will be protected by the mask layer and the sacrifice layer or the protective layer.The unprotection area is carved to effectively block the non-etching areas,accurately etch the etching area required,and solve the problem of errors.The high precision Si diffraction microlens arrays are obtained by the novel self-alignment method and the diffraction effi-ciency could reach 92.6%.After integrating with IRFPAs,the average blackbody responsity increased by 8.3%,and the average blackbody detectivity increased by 10.3%.It indicates that the Si diffraction microlens arrays can improve the filling factor and reduce crosstalk of IRFPAs through convergence,thereby improving the perfor-mance of the IRFPAs.The results are of great reference significance for improving their performance through opti-mizing the preparation level of micro nano devices.
基金Project(41174008)supported by the National Natural Science Foundation of ChinaProject(SKLGED2013-4-2-EZ)supported by the Open Foundation of State Key Laboratory of Geodesy and Earth’s Dynamics,ChinaProject(2007B51)supported by the Foundation for the Author of National Excellent Doctoral Dissertation of China
文摘The main principle and mathematical model of GOCE kinematic orbit adjustment for Earth gravity field model （EGM） validation and accelerometer calibration are presented. Based on 60 days GOCE kinematic orbits with 1-2 cm accuracy and accelerometer data from 2009-11-02 to 2009-12-31, the RMS-of-fit （ROF） of them using EGM2008, EIGEN-SC, ITG- GRACE2010S and GOCO01S up to 120, 150 and 180 degree and order （d/o） are evaluated and compared. The scale factors and biases of GOCE accelerometer data are calibrated and the energy balance method （EBM） is performed to test the accuracy of accelerometer calibration. The results show that GOCE orbits are also sensitive to EGM from 120 to 150 d/o. The ROFs of EGMs with 150 and 180 d/o are obviously better than those of EGMs with 120 d/o. The ROFs of GOCO01S and ITG-GRACE2010S are almost the same up to 120 and 150 d/o, which are about 3.3 cm and 1.8 cm, respectively. They are far better than those of EGM2008 and EIGEN-SC with the same d/o. The ROF of GOCO01S with 180 d/o is about 1.6 em, which is the best one among those EGMs. The accelerometer calibration accuracies （ACAs） of ITG-GRACE2010S and GOCO01S are obviously higher that those of EGM2008 and EIGEN-SC. The ACA of GOCO01S with 180 d/o is far higher than that of EGMs with 120 d/o, and a little higher than that of ITG-GRACE2010S with 150 d/o. I t is suggested that the newest released EGM such as GOCO01S or GOCO02S till at least 150 d/o should be chosen in GOCE precise orbit determination （POD） and accelerometer calibration.
