Abstract: Against the backdrop of digital transformation in the construction industry, video surveillance and image recognition technologies have been progressively incorporated into quality management workflows. However, conventional on-site video surveillance is inherently limited in monitoring building interiors, thereby failing to achieve comprehensive site coverage and significantly constraining its effectiveness in quality management. To overcome this limitation, this study proposes the deployment of panoramic inspection equipment to capture full-scene video of construction sites, integrated with image recognition technology for the intelligent detection and analysis of apparent quality defects, aiming to achieve more comprehensive and efficient quality management. To address the modality gap between panoramic imagery and conventional two-dimensional images, a panoramic image coordinate mapping and segmentation technique is proposed, which efficiently transforms panoramic images into a two-dimensional format compatible with existing AI models, thereby fully leveraging the enterprise's established data assets. Experimental results demonstrate that the proposed approach achieves an average accuracy of 84% in the qualitative detection of common apparent quality defects, satisfying the accuracy requirements of engineering practice. The method effectively enhances inspection efficiency and coverage while reducing omissions inherent in manual inspection, offering a scalable and innovative pathway for intelligent construction quality supervision.