Abstract: 3D laser scanning technology is applied to assist the dimensional quality inspection of houses, which can solve the problems of high labor consumption and low detection rate of traditional manual measurement methods. However, the collected point cloud data usually need to be manually segmented to extract the inspection plane, which is inefficient. Therefore, this paper proposes an engineering dimensional quality inspection method based on semantic segmentation of laser point cloud, including four aspects: point cloud data simplification, intelligent plane segmentation, intelligent semantic recognition, and dimensional quality inspection. This method effectively realizes the automatic segmentation of the point cloud data of the houses, and can automatically measure the indoor top and bottom plate elevation difference, door and window opening dimensions and other indexes based on the 3D point cloud data, which is expected to improve the efficiency and accuracy of the acceptance work of the quality of construction projects.