通过像素级XYZ坐标映射的实时6DoF姿态估计  

Real-Time 6DoF Pose Estimation via Pixel-Level XYZ Coordinates Mapping

作　　者：吴勇[1] 程良伦[1] 

机构地区：[1]广东工业大学计算机学院,广东 广州

出　　处：《计算机科学与应用》2022年第1期221-232,共12页Computer Science and Application

摘　　要：为了解决在严重遮挡和存在无纹理物体情况下,从单一RGB图像中进行6DoF姿态估计的挑战,本文提出了一种通过像素级XYZ坐标映射的实时6DoF姿态估计方法。我们引入了联合的坐标–置信度损失函数来直接回归三维模型的空间坐标,以有效地处理无纹理物体和遮挡的杂乱场景。同时,我们还考虑了2D目标检测误差导致的问题,引入了一种动态缩放策略来提高算法的性能。实验表明,我们的方法在Occlusion LINEMOD和T-LESS数据集下的评估指标优于现有的基线方法。To address the challenge of 6DoF pose estimation from a single RGB image in the presence of severe occlusion and texture-less objects, this paper proposes a real-time 6DoF pose estimation approach via pixel-level XYZ coordinates mapping. We introduce a joint coordinates-confidence loss function to directly regress the spatial coordinates of the 3D model to effectively handle texture-less objects and occluded in cluttered scenes. Meanwhile, we consider the problems caused by 2D object detection errors and introduce a dynamic scaling strategy to improve the performance of the algorithm. Experiments show that our method outperforms the existing baseline methods in terms of evaluation metrics under Occlusion LINEMOD and T-LESS datasets.

关 键 词：6DoF姿态估计 遮挡 无纹理 像素级 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]