🤖 AI Summary
Multi-view markerless 3D human pose estimation faces fundamental challenges including occlusion, 2D keypoint matching errors, arbitrary camera configurations, and scarcity of 3D ground truth annotations—hindering simultaneous achievement of accuracy, efficiency, and generalizability. This paper presents the first systematic survey of paradigm evolution in this field, critically analyzing geometric reconstruction, deep learning, graph neural networks, cross-view alignment, depth estimation, and view selection—assessing their applicability and limitations. It identifies inherent accuracy-efficiency trade-offs as a core constraint and empirically demonstrates the absence of a universally optimal solution. Methodologically, we propose a novel synergistic optimization framework integrating active learning, weak supervision, temporal consistency, and multimodal cues. Our analysis rigorously delineates performance boundaries and delivers a structured research roadmap toward real-time, high-accuracy, low-overhead practical systems.
📝 Abstract
3D human pose estimation aims to reconstruct the human skeleton of all the individuals in a scene by detecting several body joints. The creation of accurate and efficient methods is required for several real-world applications including animation, human-robot interaction, surveillance systems or sports, among many others. However, several obstacles such as occlusions, random camera perspectives, or the scarcity of 3D labelled data, have been hampering the models' performance and limiting their deployment in real-world scenarios. The higher availability of cameras has led researchers to explore multi-view solutions due to the advantage of being able to exploit different perspectives to reconstruct the pose. Thus, the goal of this survey is to present an overview of the methodologies used to estimate the 3D pose in multi-view settings, understand what were the strategies found to address the various challenges and also, identify their limitations. Based on the reviewed articles, it was possible to find that no method is yet capable of solving all the challenges associated with the reconstruction of the 3D pose. Due to the existing trade-off between complexity and performance, the best method depends on the application scenario. Therefore, further research is still required to develop an approach capable of quickly inferring a highly accurate 3D pose with bearable computation cost. To this goal, techniques such as active learning, methods that learn with a low level of supervision, the incorporation of temporal consistency, view selection, estimation of depth information and multi-modal approaches might be interesting strategies to keep in mind when developing a new methodology to solve this task.