Although significant improvement has been achieved recently in 3D human poseestimation, most of the previous methods only treat a single-person case. Inthis work, we firstly propose a fully learning-based, camera distance-awaretop-down approach for 3D multi-person pose estimation from a single RGB image.The pipeline of the proposed system consists of human detection, absolute 3Dhuman root localization, and root-relative 3D single-person pose estimationmodules. Our system achieves comparable results with the state-of-the-art 3Dsingle-person pose estimation models without any groundtruth information andsignificantly outperforms previous 3D multi-person pose estimation methods onpublicly available datasets. The code is available in https://github.com/mks0601/3DMPPE_ROOTNET_RELEASE ,https://github.com/mks0601/3DMPPE_POSENET_RELEASE.