Design And Implementation Of Illegal Building Identification Platform Based On Deep Learning

Since the reform and opening up,illegal buildings have existed like weeds.With the rhythm of economic development and the ineffective supervision of rural collective land,a large number of illegal land use in rural areas emerge in endlessly.Strengthening the protection of cultivated land and basic farmland and the governance of land planning can help the rapid and accurate identification and disposal of illegal buildings.Information supervision plays an important role in the management and control of land planning and law enforcement inspection.It is of great significance to improve the work efficiency of government departments and the development of smart land construction in the future.At the same time,the rise of artificial intelligence technology and the improvement of computer computing ability provide a feasible technical scheme for illegal building identification.In this paper,the monitoring image data captured by the camera at the high point of the communication tower is used as the data source.Compared with remote sensing image and UAV image data,it has the advantages of convenient acquisition method,high timeliness and low acquisition cost.Meanwhile,in order to improve the accuracy of the recognition results,this paper aims at the defects of Faster R-CNN in target detection at different scales.An optimization algorithm for identification of illegal buildings based on Faster R-CNN is constructed.Aiming at the problem of poor detection effect of small building targets at the distant location,the original Faster RCNN network structure was modified,and targets at different scales were detected by using low level features and high level features respectively.The experimental results show that the average accuracy of the proposed method is 75.58% in building recognition under the high point monitoring image data.Based on the real-time high point image data,the platform identifies the illegal buildings in various regions.Based on the recognition results,it builds an illegal building recognition platform to manage the illegal buildings in the whole life cycle,such as alarm research and judgment,on-site disposal,supervision and monitoring.In this paper,through the analysis and induction of the overall needs of users,the overall architecture and database of the platform are designed,and the business logic development of login module,algorithm model,alarm research and judgment,supervision and monitoring,application analysis,user management and other modules are completed.Finally,after strict functional testing,the system function meets the application requirements,and the performance test achieves the expected.