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
DOI: 10.12677/CSA.2022.121023, PDF,   
作者: 吴 勇, 程良伦：广东工业大学计算机学院，广东 广州
关键词: 6DoF姿态估计；遮挡；无纹理；像素级；6DoF Pose Estimation； Occlusion； Texture-Less； Pixel-Level
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。
Abstract: 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.
文章引用：吴勇, 程良伦. 通过像素级XYZ坐标映射的实时6DoF姿态估计[J]. 计算机科学与应用, 2022, 12(1): 221-232. https://doi.org/10.12677/CSA.2022.121023

参考文献

[1] Du, G.G., Wang, K. and Lian, S.G. (2020) Vision-Based Robotic Grasping from Object Localization, Pose Estimation, Grasp Detection to Motion Planning: A Review. Artificial Intelligence Review, 54, 1677-1734. [Google Scholar] [CrossRef] 
[2] Mur-Artal, R., Montiel, J.M.M., Tardos, J.D. (2015) ORB-SLAM: A Versatile and Accurate Monocular SLAM System. IEEE Transactions on Robotics, 31, 1147-1163. [Google Scholar] [CrossRef] 
[3] Hinterstoisser, S., Lepetit, V., Ilic, S., et al. (2012) Model Based Training, Detection and Pose Estimation of Texture-Less 3D Objects in Heavily Cluttered Scenes. Proceedings of the Asian Conference on Computer Vision, Daejeon, 5-9 November 2012, 593-596. [Google Scholar] [CrossRef] 
[4] Xiang, Y., Schmidt, T., Narayanan, V., et al. (2017) PoseCNN: A Convolutional Neural Network for 6D Object Pose Estimation in Cluttered Scenes. [Google Scholar] [CrossRef] 
[5] Tekin, B., Sinha, S.N. and Fua, P. (2018) Real-Time Seamless Single Shot 6D Object Pose Prediction. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 18-23 June 2018, 292-301. [Google Scholar] [CrossRef] 
[6] Peng. S., Liu. Y., Huang, Q., et al. (2019) PVNet: Pixel-Wise Voting Network for 6DoF Pose Estimation. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, 15-20 June 2019, 4556-4565. [Google Scholar] [CrossRef] 
[7] Zakharov, S., Shugurov, I. and Ilic, S. (2019) DPOD: 6D pose Object Detector and Refiner. 2019 IEEE/CVF International Conference on Computer Vision (ICCV), Seoul, 27 Octo-ber-2 November 2019, 1941-1950. [Google Scholar] [CrossRef] 
[8] Sundermeyer, M., Marton, Z.-C., Durner, M., et al. (2018) Implicit 3D Orientation Learning for 6D Object Detection from RGB Images. European Conference on Computer Vision (ECCV), Munich, 8-14 September 2018, 712-729. [Google Scholar] [CrossRef] 
[9] Denninger, M., Sundermeyer, M., Winkelbauer, D., et al. (2019) BlenderProc.
https://arxiv.org/abs/1911.01911
[10] Brachmann, E., Krull, A., Michel, F., et al. (2014) Learning 6D Object Pose Estimation Using 3D Object Coordinates. European Conference on Computer Vision (ECCV), Zurich, 6-12 September 2014, 536-551. [Google Scholar] [CrossRef] 
[11] Hodan, T., Haluza, P., Obdržálek, Š., et al. (2017) T-LESS: An RGB-D Dataset for 6D Pose Estimation of Texture-Less Objects. 2017 IEEE Winter Conference on Applications of Computer Vision (WACV), Santa Rosa, 24-31 March 2017, 880-888. [Google Scholar] [CrossRef] 
[12] Hodaň, T., et al. (2020) BOP Challenge 2020 on 6D Object Lo-calization. European Conference on Computer Vision (ECCV), Glasgow, 23-28 August 2020, 577-594. [Google Scholar] [CrossRef] 
[13] Oberweger, M., Rad, M. and Lepetit, V. (2018) Making Deep Heatmaps Robust to Partial Occlusions for 3D Object Pose Estimation. European Conference on Computer Vision, Munich, 8-14 September 2018, 125-141. [Google Scholar] [CrossRef] 
[14] Hodan, T., Barath, D. and Matas, J. (2020) EPOS: Estimating 6D Pose of Objects with Symmetries. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, 13-19 June 2020, 11700-11709. [Google Scholar] [CrossRef]