Research And Development Of A House Transaction Recommendation System Based On Image Enhancement Of Indoor Scenes

With the development of society,the mobility of personnel has increased,and the situation of renting houses and real estate transactions through the Internet is increasing rapidly.Some traditional online house leasing and selling systems do not optimize the indoor scenes of uploaded house pictures,and the pictures are not clear,which seriously affects the display of houses on the network platform;in addition,some network platforms do not solve the data sparseness in the recommendation system.Problems,leading to poor recommendation results.Based on the above-mentioned problems,this article researches and develops a house transaction recommendation system based on image enhancement of indoor scenes through the investigation and analysis of the current rental industry.The main work of this paper is as follows:(1)A small sample FRDSP-YOLOv3(Few-show Retinex Dark Net based on S3Pool)indoor scene recognition is proposed.For the background interference problem,a random space pooling(S3POOL)down-sampling method is adopted to preserve the spatial information of the feature map;the Retinex image enhancement method is used to solve the indoor scene target caused by factors such as the angle of the device and the light.Poor recognition improves the ability to recognize indoor scenes.Experimental results show that the accuracy rate in the indoor scene recognition task is 93%,and the recall rate is 80%.(2)In order to solve data sparsity problem in traditional collaborative algorithms,this paper proposes a user B-CF recommendation algorithm,which considers the click through rate,user browsing time,purchase records,browsing records and evaluation behavior.Using user clustering method,we first calculate the nearest neighbor users in the cluster,and then according to the score of the nearest neighbor users,we predict the items that the user does not score,which solves the problem of low accuracy caused by sparse data.(3)Designed and developed an architecture scheme of a house transaction recommendation system based on image enhancement of indoor scenes.The system uses Spring Boot and Vue framework for design and implementation.The system consists of a user website and administrator back-end services.The user website includes modules such as housing search,housing release and complaint.The back-end services implement modules such as user management,rental management,house sales management,and complaint management.