This page provides qualitative results illustrating our work on Monocular SLAM Densification. In this example, we build on ORB-SLAM 3 which provides precise camera pose estimation and triangulate sparse depth with metric scale. For the densification, we leverage the ZeroDepth model to predict a dense depth map for every keyframe. We assume that this DNN predicts depth up to a global factor, and use the SLAM outputs to correct the scale. Lastly, we integrate Voxblox to build a voxel map iteratively from the resulting scaled dense depth maps and their corresponding estimated camera pose. These experiments were carried out on the EuRoC V1_01 sequence and we displayed the 3D scan of the room in white. Further details are provided in the publishes papers (see Publications).