Starting from the characteristics of modernization in three different historical stages since the founding of the People’s Republic of China,this paper analyses the internal changes in the professional name,subordina...Starting from the characteristics of modernization in three different historical stages since the founding of the People’s Republic of China,this paper analyses the internal changes in the professional name,subordinate disciplines and categories,and professional mission of the environmental design profession,and summarizes the changes in the characteristics of the environmental design profession in 10 aspects such as subordinate disciplines,design objects,design types,cultural characteristics,and aesthetic values on the basis of the results of the analysis,and explores five transformations of the environmental design profession in the process of China’s development in the future.The five transformations in the future development of China are to serve the harmony and beauty of the three major urban and rural spaces with“small and micro-renewal and reconstruction”;to serve the transmission of Chinese culture and the spread of Chinese civilization with“spatial heritage and innovation”;to serve the balance and sufficiency of the four functions of urban and rural settlements with“friendliness and equilibrium”;to serve the intelligence,wisdom,and enjoyment of indoor and outdoor living environment space with“scientific and technological achievements”;and to serve the co-construction,co-management,and sharing of indoor and outdoor public space in urban and rural areas with“public participation.”展开更多
针对无人机场景下行人重识别所呈现的多视角多尺度特点,以及传统的基于卷积神经网络的行人重识别算法受限于局部感受野结构和下采样操作,很难对行人图像的全局特征进行提取且图像空间特征分辨率不高。提出一种无人机场景下基于Transfor...针对无人机场景下行人重识别所呈现的多视角多尺度特点,以及传统的基于卷积神经网络的行人重识别算法受限于局部感受野结构和下采样操作,很难对行人图像的全局特征进行提取且图像空间特征分辨率不高。提出一种无人机场景下基于Transformer的轻量化行人重识别(Lightweight Transformer-based Person Re-Identification,LTReID)算法,利用多头多注意力机制从全局角度提取人体不同部分特征,使用Circle损失和边界样本挖掘损失,以提高图像特征提取和细粒度图像检索性能,并利用快速掩码搜索剪枝算法对Transformer模型进行训练后轻量化,以提高模型的无人机平台部署能力。更进一步,提出一种可学习的面向无人机场景的空间信息嵌入,在训练过程中通过学习获得优化的非视觉信息,以提取无人机多视角下行人的不变特征,提升行人特征识别的鲁棒性。最后,在实际的无人机行人重识别数据库中,讨论了在不同量级主干网和不同剪枝率情况下所提LTReID算法的行人重识别性能,并与多种行人重识别算法进行了性能对比,结果表明了所提算法的有效性和优越性。展开更多
基金Chongqing 2023 Undergraduate Colleges and Universities“Course Ideology and Politics Demonstration Course”and First-Class Undergraduate Course“Offline Course”“Environmental Space Design,”Chongqing College of Engineering 2022 Undergraduate“Course Civics and Politics Demonstration Course”and Gold Course“Offline Course”“Environmental Space Design,”Chongqing 2021 Higher Education Teaching Reform Research Project-Teaching Reform and Practice of Curriculum Civics and Politics Education Integrated into Environmental Design Professional Courses(Key Project)(202127)。
文摘Starting from the characteristics of modernization in three different historical stages since the founding of the People’s Republic of China,this paper analyses the internal changes in the professional name,subordinate disciplines and categories,and professional mission of the environmental design profession,and summarizes the changes in the characteristics of the environmental design profession in 10 aspects such as subordinate disciplines,design objects,design types,cultural characteristics,and aesthetic values on the basis of the results of the analysis,and explores five transformations of the environmental design profession in the process of China’s development in the future.The five transformations in the future development of China are to serve the harmony and beauty of the three major urban and rural spaces with“small and micro-renewal and reconstruction”;to serve the transmission of Chinese culture and the spread of Chinese civilization with“spatial heritage and innovation”;to serve the balance and sufficiency of the four functions of urban and rural settlements with“friendliness and equilibrium”;to serve the intelligence,wisdom,and enjoyment of indoor and outdoor living environment space with“scientific and technological achievements”;and to serve the co-construction,co-management,and sharing of indoor and outdoor public space in urban and rural areas with“public participation.”
文摘针对无人机场景下行人重识别所呈现的多视角多尺度特点,以及传统的基于卷积神经网络的行人重识别算法受限于局部感受野结构和下采样操作,很难对行人图像的全局特征进行提取且图像空间特征分辨率不高。提出一种无人机场景下基于Transformer的轻量化行人重识别(Lightweight Transformer-based Person Re-Identification,LTReID)算法,利用多头多注意力机制从全局角度提取人体不同部分特征,使用Circle损失和边界样本挖掘损失,以提高图像特征提取和细粒度图像检索性能,并利用快速掩码搜索剪枝算法对Transformer模型进行训练后轻量化,以提高模型的无人机平台部署能力。更进一步,提出一种可学习的面向无人机场景的空间信息嵌入,在训练过程中通过学习获得优化的非视觉信息,以提取无人机多视角下行人的不变特征,提升行人特征识别的鲁棒性。最后,在实际的无人机行人重识别数据库中,讨论了在不同量级主干网和不同剪枝率情况下所提LTReID算法的行人重识别性能,并与多种行人重识别算法进行了性能对比,结果表明了所提算法的有效性和优越性。