Application and Optimization of AI-Driven Point Cloud Classification Algorithms in Landscape Architecture

With the widespread application of three-dimensional point cloud technology in urban environment modeling and smart landscape management, how to achieve high-precision object recognition in complex landscape scenes has become an urgent problem to be solved. This paper proposes an artificial intelligence-based point cloud classification algorithm, which integrates local feature enhancement and multi-scale context modeling mechanism, and significantly improves the model's ability in boundary recognition and semantic consistency. Based on the PointNet++ architecture, the proposed method introduces the attention mechanism weighted neighborhood features, and effectively models complex spatial structures through multi-scale feature aggregation. Experimental results on the Suzhou Classical Gardens and Hangzhou West Lake real-time data sets show that the proposed method is superior to mainstream algorithms in terms of overall accuracy and average intersection-over-union ratio, and has strong generalization ability and practical application value. The research results provide strong technical support for landscape digital modeling, ecological protection and smart scenic area construction.