首页|风景园林图像与图形在深度学习中的应用分析及未来展望

风景园林图像与图形在深度学习中的应用分析及未来展望

扫码查看
深度学习处理图像与图形数据的广泛应用,为风景园林研究的大样本数据获取、分析、预测,以及景观设计图的快速生成提供了新的解决思路与有效途径。本文以风景园林图像与图形为研究对象,剖析风景园林图像与图形的类型,探究其在深度学习技术中的应用途径,分别从图像识别、图像生成、图形预测三个方面出发,对国内外的相关文献进行分析总结,梳理应用进展,提出未来发展趋势可聚焦深度学习向迁移学习的转变、人工智能与创意思维的融合、物质属性与非物质属性的结合,并强调深度学习技术通过处理风景园林图像与图形在分析场所空间环境、自动生成景观表现图、快速智能化建模、科学预判人群行为偏好等方面发挥着巨大的作用,将其应用于风景园林领域,能够有效推动本学科的智慧化发展。
Application Analysis and Future Prospect of Landscape Architecture Images and Graphics in Deep Learning
New approaches and efficient methods for the collection,analysis,and forecasting of big sample data in landscape architecture research,as well as the quick creation of landscape design drawings,are made possible by the widespread use of deep learning to handle image and graphic data.This paper took landscape architecture images and graphics as its object,analyzed the types of landscape architecture images and graphics,and explored their application ways in deep learning technology.Starting from three aspects of image recognition,image generation,and graphic prediction,this paper analyzed and summarized relevant literatures at home and abroad,combed the application progress,and proposed that the future development trend can focus on the transformation from deep learning to transfer learning,the integration of artificial intelligence and creative thinking,the combination of material attributes and non-material attributes,and emphasized the significance of deep learning technology in analyzing the space environment of location,automatically generating landscape representation maps,rapid intelligent modeling,scientific prediction of crowd behavior preferences,and so on,by processing landscape architecture images and graphics.Its application in landscape architecture can effectively promote the intelligent development of this profession.

deep learninglandscape images and graphicsimage recognitionimage generationgraph prediction

刘冠、邵继中、王宇琪、张雪茵、吕欣蓓

展开 >

华中农业大学园艺林学学院,湖北 武汉 430070

深度学习 风景园林图像与图形 图像识别 图像生成 图形预测

国家重点研发计划住建部科学技术计划国际科技合作类项目

2023YFC38075002022-H-001

2024

南京师大学报(自然科学版)
南京师范大学

南京师大学报(自然科学版)

CSTPCD北大核心
影响因子:0.427
ISSN:1001-4616
年,卷(期):2024.47(2)
  • 56