At present， construction visualization technology based on mobile Augmented Reality （AR） has the advantage of diversified interaction methods， but it has disadvantages of low accuracy and poor system integration， making it difficult to be applied to in-depth applications such as progress investment statistics. To address the above problem， based on the Simultaneous Localization And Mapping （SLAM） technology， real-time and all-round images of the construction site were collected. At the same time， Building Information Model （BIM） and image recognition technology were introduced to achieve real-time registration between on-site construction scenes and BIMs based on mobile AR technology. Engineering quantity comparison calculations were carried out to analyze the actual completion of the project， monitor the project progress accurately， verifying the practical effect of the proposed method in engineering.