A visual simultaneous localization and mapping(vSLAM) system in a dynamic environment are affected by the wrong associated data caused by the moving targets, which causes a large error in the pose estimation of the mobile robot and affects the subsequent tasks of the robot. Combining semantic segmentation information to remove dynamic feature points in the image is an effective method to improve the accuracy of the SLAM system. However, the existing visual SLAM based on semantic segmentation usually adopts the fully supervised approaches to segment the dynamic scenes, which depends on a large number of training data sets with labelled information to guarantee accuracy and limits the application of SLAM system. To address this issue, a visual semantic SLAM system(vsSLAM) that applies weakly supervised semantic segmentation to dynamic scenes is proposed to broaden the application range of the system. Firstly, the system extracts the features of the input image and checks the moving consistency, and then segments the dynamic target with the weakly supervised methods. Secondly, the semantic segmentation results are used to remove the dynamic feature points in the image. Finally，the stable static feature points are adopted to carry out the pose estimation. Experiments were performed on the public TUM data sets. The results show that the accuracy of the SLAM system based on the weakly supervised network adopted in this paper is significantly higher than that of the traditional ORB-SLAM2 system, and also higher than the SLAM system of the weakly supervised network SEC. The accuracy is close to the fully supervised semantic SLAM system.