Research On Urban Wealth Perception Based On Baidu Street View

The research on urban rich and poor space has always been a hot topic in the research of urban issues.The process of urbanization is affected by many factors such as the natural environment,historical development,technological changes,policy adjustments,etc.A variety of complex factors cause the urban construction and economic development will inevitably appear the phenomenon of internal imbalance.Studying urban wealth perception and exploring its distribution characteristics is conducive to urban planning and related policy formulation.However,urban wealth perception is a subjective concept.Existing research methods are mostly based on the analysis of urban residents' income,housing prices and other economic indicators related to urban wealth perception.Therefore,the acquisition of research data depends on the traditional survey sampling method,which has high cost,long time and poor scalability.As time goes by,urban development requires re-investigation and research.This article will use street view and deep learning technology to build a model that automatically predicts urban wealth perception based on street view,and verify the rationality of the model with relevant data.The work done in this article is mainly reflected in the following aspects:(1)Based on the crawler technology,the street view images of the street attractions in the main urban area of Nanchang were obtained,and a street view image data set with balanced categories of the research area was constructed.Aiming at the problem of skewness of labeled data,this paper proposes a spatial sampling method to down-sampling too many categories and over-sampling too few categories to get a balanced data set.(2)Build a model of urban wealth perception based on deep learning.The model takes8 images of street spots as input,obtains the high-level features of the image through the semantic segmentation model,then trains and categorizes the features of street spots,finally obtains the wealth perception level of each street spot.Experimental results show that the accuracy of the model reached 74.83%.(3)The relationship between the wealth perception in the experimental area and the road grade,POI,and housing price data is studied.The experimental results show that: 1)Except for expressways and expressways,the wealth perception is generally lower,the higher the road level,the better the wealth perception;2)POI areas such as shopping malls,movie theaters,and subway stations all have a better wealth perception,college POI's overall wealth perception in the region is the lowest;3)The analysis of wealth perception and housing prices found that for the entire study area,the correlation between housing prices and wealth perception is low,but in some areas,they have a significant positive correlation,namely In some local areas,the higher the housing price of the community,the better the wealth perception in the surrounding area.