Font Size: a A A

Deep Learning Model-based Analysis Of Forest Landscape Preferences And Emotional Features

Posted on:2024-06-09Degree:MasterType:Thesis
Country:ChinaCandidate:X T ZengFull Text:PDF
GTID:2543306938489064Subject:Forest science
Abstract/Summary:PDF Full Text Request
Forests provide a variety of functions for humans around the world,and in the process of participating in forest recreation,people will generally prefer the main landscape element in the forest landscape-plants,whose aesthetic role is the most dominant and the first to be recognized and used.From the experience of forest tourism use it is concluded that the forest landscape is the total component of the visual image of the forest.The study of forest landscape preference has an important role and significance for forest landscape conservation,quality improvement and utilization.The advent of the Web 2.0 era will provide more effective research data for forest landscape preference research.The field of artificial intelligence including computer vision(CV),image processing techniques,and deep learning have made significant progress,providing many new possibilities and ideas for the study of preferences as well as tourism photos.Based on the review of related concepts and researches at home and abroad,this study selects six case sites in China based on deep learning technology,and takes the geotagged photos of forest landscape posted by forest recreationists on the "Two Steps"app as the research object.The preferences of eight forest landscape scenes,including Look Down Landscape,Look Forward Landscape,Look Up Landscape,Single-tree Composed Landscape,Detailed Landscape,Overall Landscape,Forest Trail Landscape and Intra-forest Landscape,were explored from three dimensions,including Patial Hierarchy,Forest Hierarchy,and Scale Hierarchy.Meanwhile,this study processed the forest landscape photos by Deepsentibank tool and used Hownet sentiment lexicon and Russell sentiment loop model to perform sentiment analysis on the obtained results to better understand the forest landscape preferences of recreationists.The final research results show that:(1)From the aesthetic spatial angle,people prefer flat view,that is,the eye height as the base point of the starting visual threshold range of forest landscape.The attention of the upward and downward views is relatively low,and recreationists generally do not look up or down to view the landscape.Different forest landscape species,such as different seasons will also appear to be a higher preference for Look Up Landscape than Look Down Landscape.(2)From the forest,scale hierarchy,forest trail landscape has a high preference,implying that trail landscape plays an important role on forest landscape creation,and the landscape inside the forest has a certain preference.but the preference of single-tree composed,detailed,and overall landscape is low,which proves that recreationists pay little attention to flowers,fruits,and individual combination of trees inside the forest,and look more at the local landscape inside the forest.(3)Forest landscape photographs are taken although positive vocabulary and positive emotional states are extremely high,but negative emotions still exist.Forest recreationists’ emotions are a complex state of positive,negative or both interactions rather than a single positive emotion,thus showing from the side that people’s preferences can have both positive and negative preferences.(4)Forest recreationists have a clear time preference when carrying out forest activities.In terms of visitation time periods,they are mostly concentrated in the daytime period,with the peak of forest recreation activities in October.Seasonally,there is a preference for the colorful fall and the better summer months.
Keywords/Search Tags:Forest landscape, Deepsentibank, Deep learning, Geotagged photos, Sentiment analysis
PDF Full Text Request
Related items