Pose-invariant facial expression recognition(FER)is an active but challenging research topic in computer vision.Especially with the involvement of diverse observation angles,FER makes the training parameter models inc...Pose-invariant facial expression recognition(FER)is an active but challenging research topic in computer vision.Especially with the involvement of diverse observation angles,FER makes the training parameter models inconsistent from one view to another.This study develops a deep global multiple-scale and local patches attention(GMS-LPA)dual-branch network for pose-invariant FER to weaken the influence of pose variation and selfocclusion on recognition accuracy.In this research,the designed GMS-LPA network contains four main parts,i.e.,the feature extraction module,the global multiple-scale(GMS)module,the local patches attention(LPA)module,and the model-level fusion model.The feature extraction module is designed to extract and normalize texture information to the same size.The GMS model can extract deep global features with different receptive fields,releasing the sensitivity of deeper convolution layers to pose-variant and self-occlusion.The LPA module is built to force the network to focus on local salient features,which can lower the effect of pose variation and self-occlusion on recognition results.Subsequently,the extracted features are fused with a model-level strategy to improve recognition accuracy.Extensive experimentswere conducted on four public databases,and the recognition results demonstrated the feasibility and validity of the proposed methods.展开更多
A circular thin plate is proposed for vibration attenuation,which is attached alternately by annular piezoelectric unimorphs with resonant shunt circuits.Two kinds of equal frequency resonant shunt circuits are design...A circular thin plate is proposed for vibration attenuation,which is attached alternately by annular piezoelectric unimorphs with resonant shunt circuits.Two kinds of equal frequency resonant shunt circuits are designed to achieve an integrated locally resonant（LR）band gap（BG） with a much smaller transmission factor：（1） the structure is arrayed periodically while the resonant shunt circuits are aperiodic;（2） the resonant shunt circuits are periodic while the structure is aperiodic.The transmission factor curve is calculated,which is validated by the finite element method.Dependences of the LR BG performance upon the geometric and electric parameters are also analyzed.展开更多
Nutrients are distributed unevenly in the soil.Phenotypic plasticity in root growth and proliferation may enable plants to cope with this variation and effectively forage for essential nutrients. However, how micronut...Nutrients are distributed unevenly in the soil.Phenotypic plasticity in root growth and proliferation may enable plants to cope with this variation and effectively forage for essential nutrients. However, how micronutrients shape root architecture of plants in their natural environments is poorly understood. We used a combination of field and laboratory-based assays to determine the capacity of Nicotiana attenuata to direct root growth towards localized nutrient patches in its native environment. Plants growing in nature displayed a particular root phenotype consisting of a single primary root and a few long, shallow lateral roots. Analysis of bulk soil surrounding the lateral roots revealed a strong positive correlation between lateral root placement and micronutrient gradients, including copper, iron and zinc. In laboratory assays, the application of localized micronutrient salts close to lateral root tips led to roots bending in the direction of copper and iron. This form of chemotropism was absent in ethylene and jasmonic acid deficient lines,suggesting that it is controlled in part by these two hormones. This work demonstrates that directed root growth underlies foraging behavior, and suggests that chemotropism and micronutrient-guided root placement are important factors that shape root architecture in nature.展开更多
Benefiting from the advancement of deep learning techniques,face photo-sketch synthesis has witnessed significant progress in recent years.Cutting-edge methods typically treat this task as an image-to-image translatio...Benefiting from the advancement of deep learning techniques,face photo-sketch synthesis has witnessed significant progress in recent years.Cutting-edge methods typically treat this task as an image-to-image translation problem and train a conditional generative model to learn the mapping between two domains.However,purely parametric deep learning models often struggle to capture instance-level details due to limited training samples and tend to focus on domain-level mapping.Moreover,sketch-to-photo synthesis is more challenging than photo-to-sketch synthesis and holds greater significance in the realm of public security,but it has not been well-studied in existing methods.To address these challenges,we introduce an innovative framework that synergistically integrates parametric and non-parametric approaches,infusing facial generative priors and instancelevel prior knowledge from the target domain to enrich texture detail synthesis.Specifically,our framework employs a semantic-aware network to facilitate coarse cross-domain reconstruction,thereby capturing domain-level information.Moreover,through efficient neural patch matching between the input image and multiple reference(training)samples,we can harness instance-level prior knowledge as a detailed texture representation to enhance detail fidelity.For the sketch-to-photo synthesis task,we further propose a local patch correspondence mechanism that improves the rationality of matching through local constraint.To further enhance the generation of realistic and detailed facial features,we incorporate a pre-trained StyleGAN as the decoder,leveraging its extensive facial generative priors.Additionally,we introduce the relaxed Earth Movers Distance(rEMD)loss to improve the style consistency between the generated results and the target domain.Extensive experiments show that our method achieves state-of-the-art performance on both quantitative and qualitative evaluations.展开更多
基金supported by the National Natural Science Foundation of China (No.31872399)Advantage Discipline Construction Project (PAPD,No.6-2018)of Jiangsu University。
文摘Pose-invariant facial expression recognition(FER)is an active but challenging research topic in computer vision.Especially with the involvement of diverse observation angles,FER makes the training parameter models inconsistent from one view to another.This study develops a deep global multiple-scale and local patches attention(GMS-LPA)dual-branch network for pose-invariant FER to weaken the influence of pose variation and selfocclusion on recognition accuracy.In this research,the designed GMS-LPA network contains four main parts,i.e.,the feature extraction module,the global multiple-scale(GMS)module,the local patches attention(LPA)module,and the model-level fusion model.The feature extraction module is designed to extract and normalize texture information to the same size.The GMS model can extract deep global features with different receptive fields,releasing the sensitivity of deeper convolution layers to pose-variant and self-occlusion.The LPA module is built to force the network to focus on local salient features,which can lower the effect of pose variation and self-occlusion on recognition results.Subsequently,the extracted features are fused with a model-level strategy to improve recognition accuracy.Extensive experimentswere conducted on four public databases,and the recognition results demonstrated the feasibility and validity of the proposed methods.
基金Project supported by the National Natural Science Foundation of China(Nos.11272126,51435006,and 51421062)the Fundamental Research Funds for the Central Universities,HUST:2016JCTD114 and 2015TS121the Research Fund for the Doctoral Program of Higher Education of China(No.20110142120050)
文摘A circular thin plate is proposed for vibration attenuation,which is attached alternately by annular piezoelectric unimorphs with resonant shunt circuits.Two kinds of equal frequency resonant shunt circuits are designed to achieve an integrated locally resonant（LR）band gap（BG） with a much smaller transmission factor：（1） the structure is arrayed periodically while the resonant shunt circuits are aperiodic;（2） the resonant shunt circuits are periodic while the structure is aperiodic.The transmission factor curve is calculated,which is validated by the finite element method.Dependences of the LR BG performance upon the geometric and electric parameters are also analyzed.
基金the Max Planck Society,European Research Council Advanced Grant(No.293926 to ITB)Alexander von Humboldt Postdoctoral Research Fellowship (to APF)+1 种基金Brazilian National Council for Research (CNPq grant no.237929/2012-0 to CCMA) for financial support
文摘Nutrients are distributed unevenly in the soil.Phenotypic plasticity in root growth and proliferation may enable plants to cope with this variation and effectively forage for essential nutrients. However, how micronutrients shape root architecture of plants in their natural environments is poorly understood. We used a combination of field and laboratory-based assays to determine the capacity of Nicotiana attenuata to direct root growth towards localized nutrient patches in its native environment. Plants growing in nature displayed a particular root phenotype consisting of a single primary root and a few long, shallow lateral roots. Analysis of bulk soil surrounding the lateral roots revealed a strong positive correlation between lateral root placement and micronutrient gradients, including copper, iron and zinc. In laboratory assays, the application of localized micronutrient salts close to lateral root tips led to roots bending in the direction of copper and iron. This form of chemotropism was absent in ethylene and jasmonic acid deficient lines,suggesting that it is controlled in part by these two hormones. This work demonstrates that directed root growth underlies foraging behavior, and suggests that chemotropism and micronutrient-guided root placement are important factors that shape root architecture in nature.
基金supported in part by the National Natural Science Foundation of China(U22A2096 and 62441601)in part by the Shaanxi Province Core Technology Research and Development Project(2024QY2-GJHX-11)+1 种基金in part by the Fundamental Research Funds for the Central Universities(QTZX23042)in part by the Innovation Fund of Xidian University(YJSJ24017).
文摘Benefiting from the advancement of deep learning techniques,face photo-sketch synthesis has witnessed significant progress in recent years.Cutting-edge methods typically treat this task as an image-to-image translation problem and train a conditional generative model to learn the mapping between two domains.However,purely parametric deep learning models often struggle to capture instance-level details due to limited training samples and tend to focus on domain-level mapping.Moreover,sketch-to-photo synthesis is more challenging than photo-to-sketch synthesis and holds greater significance in the realm of public security,but it has not been well-studied in existing methods.To address these challenges,we introduce an innovative framework that synergistically integrates parametric and non-parametric approaches,infusing facial generative priors and instancelevel prior knowledge from the target domain to enrich texture detail synthesis.Specifically,our framework employs a semantic-aware network to facilitate coarse cross-domain reconstruction,thereby capturing domain-level information.Moreover,through efficient neural patch matching between the input image and multiple reference(training)samples,we can harness instance-level prior knowledge as a detailed texture representation to enhance detail fidelity.For the sketch-to-photo synthesis task,we further propose a local patch correspondence mechanism that improves the rationality of matching through local constraint.To further enhance the generation of realistic and detailed facial features,we incorporate a pre-trained StyleGAN as the decoder,leveraging its extensive facial generative priors.Additionally,we introduce the relaxed Earth Movers Distance(rEMD)loss to improve the style consistency between the generated results and the target domain.Extensive experiments show that our method achieves state-of-the-art performance on both quantitative and qualitative evaluations.
