基于深度学习的建筑物识别

针对随着城市化的快速发展,城市与城市间的辨识度越来越弱,城市地标的概念越来越热门这一现象,提出了一种基于深度学习的建筑物识别方法;使用改进的Faster R-CNN算法作为训练模型,首先,将需要识别的图片输入深度神经网络,提取出特征框图;然后,模型通过区域建议网络预测目标建筑物所在位置的区域建议,并将这些区域建议映射到特征框图上,RoI Pooling层将这些区域建议转换成固定大小的特征框图;最后使用非极大值抑制从预测边界框中移除相似的结果,得到预测边界框以及边框中目标建筑物的类别和概率;实验结果表明:在训练数据集充足的条件下,使用此方法对地标建筑物的识别率能达到90. 8%,通过与其他模型比较分析,该模型具有较好的识别效果。     

新疆两河源自然保护区鼓藻属植物分布新资料

报道了采自新疆两河源国家级自然保护区的鼓藻属(绿藻门)新疆新记录植物9种，包括3种5变种1变型，分别为厚皮鼓藻增厚变种C. pachydermum var. incrassatum Scottet et Gr?nblad、圆齿鼓藻圆齿变种博尔变型C. crenatum var. crenatum f. boldtiana (Gutwinski) West & West、帽状鼓藻C. galeretum Nordsted、近缘鼓藻C. connatum Ralfs、穿孔纹鼓藻斯科亚变种C. perforatum var. skujae Gr?nblad、方形鼓藻C. quadratum Ralfs、亚脊鼓藻贝克变种C. subcostatum var. beckii (Gutwinski) West & West、四眼鼓藻C. tetraophthalmum Ralfs、特平鼓藻拔萃变种C. turpinii var.eximium West & West.对每一个种(包括变种及变型)的形态分类学特征进行了详细的描述，编制了检索表，丰富了新疆鼓藻类植物研究资料.     

A new superfamily,family, genus and species of marine amphipod,Protodulichia scandens,from Japan (Crustacea: Amphipoda: Senticaudata: Corophiida)

ABSTRACT

In this study, the new amphipod superfamily Protodulichioidea and family Protodulichiidae are established. Both belong to the infraorder Corophiida but is different from all known superfamilies and families. Amphipods included in the new superfamily and family are characterised by having a triangular head with large eyes, extremely long pereopod 7, and uropod 3 with a short peduncle and slender rami bearing a few robust setae. Protodulichia scandens gen. et sp. nov. is fully described here and its mast-building behaviour is also recorded.

urn:lsid:zoobank.org:pub:F6613B27-D954-4E4C-86CB-9FC5E9624334     

新型城镇化背景下绿色交通的发展对策

 通过分析绿色交通当前的发展态势，结合国内外发展的实际情况以及新型城镇化发展的要求，从宏观、中观、微观3个层面分析了绿色交通发展面临的挑战。提出绿色交通的发展应该关注各城市、各地区的差异化特征，并需要在不断应对新技术挑战的基础上，全面提升交通与用地布局的协调性，同时应全力优化交通结构，关注细节，建立"公共交通+慢行"主导的交通模式。     

New data on sponges from Svalbard Archipelago with a description of a new species of Halicnemia

Sponge fauna from the west and south of Svalbard archipelago was studied. A total of 28 species of sponges were identified. Five of them are new records to the study area and one is new to science. New records for Spitsbergen are: Axinella hispida (Koltun 1959), Haliclona rossica (Hentschel 1929), Myxilla (Myxilla) perspinosa (Lundbeck 1905), Sphaerotylus borealis (Swartschewsky 1906) and Suberites carnosus (Johnston 1842). A new species of sponge, Halicnemia wagini sp. nov., collected south-west of Spitsbergen on the continental slope at a depth of 423–425 m, is described. It is similar to Halicnemia patera (Bowerbank 1864) and H. verticillata (Bowerbank 1866) in skeleton architecture, but significantly differs in its spicule complement and external appearance. Primarily, distinct differences concern the morphology of microscleres. In comparison with Halicnemia patera, the new species has an additional category of acanthose microscleres presented by microstrongyles, while microscleres of H. verticillata are only slightly curved and also verticillately spined. The study area, west of Spitsbergen, establishes close relationships with North Atlantic due to warm Atlantic waters of the West Spitsbergen Current. Proximity of the new species to the North Atlantic representatives of Halicnemia genus allowed us to suggest that H. wagini sp. nov., both ecologically and morphologically, has stood apart as a distinct biological species in the conditions of the Arctic.

http://www.zoobank.org/lsid:zoobank.org:pub:2A09537B-63D7-4F07-A2AB-9C707A686333     

河南省新型农村社区建设的动力机制研究

为缓解城乡发展差距,推进农村的发展,中央先后提出"推进社会主义新农村建设"和"积极推进农村社区建设",通过完善农村的政治、经济、文化、社会保障等方面,达到城乡发展一体化。河南省作为农业大省,农村的发展显得尤为重要,但在农村发展过程中依然存在动力不足的问题,在中原经济区建设的背景下探讨河南省农村发展的动力问题则显得更有意义。     

四螺旋创新集群:研究型大学人工智能发展生态重构与路向探究——以加拿大多伦多大学为例

以人工智能为代表的知识迭代发展,促使知识生产模式发生重大转变,催生了“大学—产业—政府—公众”四螺旋动力结构的建立和研究型大学知识生产新生态重构。以多伦多大学人工智能发展为例,考察其四螺旋利益相关主体及实践。从内在机理看,4个主体在互动运作中不仅重新确定了各自角色,而且还建立了大学(知识)—产业(产品)—政府(治理)—公众(公益)的新型逻辑链条,平衡公私利益格局,把公益指向作为人工智能四螺旋运作的中心目标;从外在特征看,4个主体形成了以大学为中心的区域创新网络,并牵引其他主体形成环高校创新集群。当前,我国在人工智能发展中不断发力,已进入世界第一方阵,但仍需要在国际比较学习中不断提升自身竞争力。基于案例分析,我国研究型大学发展人工智能有必要把握4个方面:一是走进中心,塑造大学在人工智能发展中的领导地位;二是以专促通,创新研究型大学人工智能专业培养模式;三是引企入研,提升校企合作人工智能创新的转化升级;四是人本导向,突出大学在人工智能发展中的公共价值。总体来看,我国研究型大学发展人工智能不仅要面临技术上的挑战,更要面临来自治理的挑战。