In recent years, simultaneous localization and mapping (SLAM) technologies have made significant advances in the fields of autonomous navigation and map generation in dynamic environments. Among the SLAM technologies, ORBSLAM3 has demonstrated outstanding efficiency and accuracy in relation to three-dimensional map generation and location estimation using monocular, stereo, and RGB-D cameras. However, the limitations of visual place recognition in environments with insufficient visual features or repetitive patterns continue to affect the overall performance of ORBSLAM3. In this paper, we present two methods that use QR codes as artificial landmarks to address these limitations and propose an algorithm that integrates these methods into ORBSLAM3 to improve performance. The first method uses a QR code image as a visual feature provider, and the second method improves the accuracy and speed of place recognition via an image similarity search based on label data of QR codes. This study demonstrates that this integration is an effective method to overcome the limitations of visual place recognition and improve the performance of the system.