Összehasonlítás a 2D-s és 3D-s tárgyfelismerő technikák között a robot navigációban

Comparison between 2D and 3D Object Recognition Techniques for Mobile Robot Navigation

  • MOLNÁR Szilárd
  • TAMÁS Levente
Keywords: mesterséges intelligencia, tárgyfelismerés, kaolin, YOLO, tér feltérképezés

Abstract

This is a short presentation about the comparison between a 2D object recognition technique, YOLO, and a 3D object recognition technique with Kaolin. These techniques are analyzed, and used in navigating a small robot through a hallway. After presenting the used hardware elements, we will conclude the results of the comparison according to the effectiveness of the application.

Kivonat

Ez egy rövid leírás egy projektről, amelyikben egy összehasonlítunk egy 2D-s tárgyfelismerő tech­ni­kát, a YOLO-t, és egy 3D-s tárgyfelismerő technikát, a Kaolint. Ezeknek a működését analizáljuk, majd fel­használjuk őket egy kisméretű robot folyosón való irányításánál. Ismertetjük a használt eszközöket, majd az alkalmazás hatékonyságát elemezve, levonjuk az összehasonlítás következményét.

References

M. Bjelonic, "{YOLO ROS}: Real-Time Object Detection for {ROS}," 2016. [Online]. Available: https://github.com/leggedrobotics/darknet_ros}. [Accessed 25 5 2020].

Y. a. W. H. a. H. Q. a. L. H. a. L. L. a. B. M. Guo, "Deep Learning for 3D Point Clouds: A Survey," arXiv preprint arXiv:1912.12033, 2019.

C. R. a. S. H. a. M. K. a. G. L. J. Qi, "PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation," arXiv preprint arXiv:1612.00593, 2016.

C. R. a. Y. L. a. S. H. a. G. L. J. Qi, "PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space," arXiv preprint arXiv:1706.02413, 2017.

W. a. Q. Z. a. F. L. Wu, "PointConv: Deep Convolutional Networks on 3D Point Clouds," arXiv preprint arXiv:1811.07246, 2018.

Y. a. F. T. a. X. M. a. Z. L. a. Q. Y. Xu, "SpiderCNN: Deep Learning on Point Sets with Parameterized Convolutional Filters," arXiv preprint arXiv:1803.11527, 2018.

W. a. Q. Z. a. F. L. Wu, "PointConv: Deep Convolutional Networks on 3D Point Clouds," arXiv preprint arXiv:1811.07246, 2018.

Y. a. T. O. Zhou, "PointPillars," arXiv, no. 10.1109/CVPR.2018.00472.

Y. a. M. Y. a. L. B. Yan, "SECOND : Sparsely Embedded," no. 10.3390/s18103337, pp. 1-17, 2018.

C. R. a. L. O. a. H. K. a. G. L. J. Qi, "Deep Hough Voting for 3D Object Detection in Point Clouds," 2019.

S. a. W. X. a. L. H. Shi, "PointRCNN: 3D Object Proposal Generation and Detection From Point Cloud," 2019.

A. a. C. A. X. a. S. M. a. H. M. a. F. T. a. N. M. Dai, "ScanNet: Richly-Annotated 3D Reconstructions of Indoor Scenes," 2017.

Z. a. J. K. Wang, "Frustum ConvNet: Sliding Frustums to Aggregate Local Point-Wise Features for Amodal 3D Object Detection," 2019.

T. R. 3. O. D. f. P. C. w. T. Attention, "Zhe Liu and Xin Zhao and Tengteng Huang and Ruolan Hu and Yu Zhou and Xiang Bai," AAAI, no. 1912.05163, 2020.

D. a. S. S. Maturana, "VoxNet: A 3D Convolutional Neural Network for Real-Time Object Recognition," 2015.

A. {. a. R. {. a. C. Y. {. a. R. {. a. A. {Bab-Hadiashar}, "PL-Net3d: Robust 3D Object Class Recognition Using Geometric Models," IEEE Access, vol. 7, no. 10.1109/ACCESS.2019.2952638, pp. 163757-163766, 2019.

M. a. S. E. a. L. J.-F. a. {. T. C. a. R. A. a. C. W. a. X. T. a. L. R. a. F. S. J., "Kaolin: A PyTorch Library for Accelerating 3D Deep Learning Research," arXiv:1911.05063, 2019.

"Kaolin example Classification," [Online]. Available: https://github.com/NVIDIAGameWorks/kaolin/tree/master/examples/Classification. [Accessed 03 05 2020].

"Description about PolygonMesh message type," [Online]. Available: http://docs.ros.org/fuerte/api/pcl/html/msg/PolygonMesh.html. [Accessed 03 05 2020].

"ModelNet10 Dataset source," [Online]. Available: https://modelnet.cs.princeton.edu/. [Accessed 03 05 2020].

"Description about the .off file format," [Online]. Available: https://shape.cs.princeton.edu/benchmark/documentation/off_format.html. [Accessed 03 05 2020].

"Description about the .pcd file format," [Online]. Available: https://www.online-convert.com/file-format/pcd. [Accessed 03 05 2020].

"Description about the .vtk file format," [Online]. Available: https://vtk.org/wp-content/uploads/2015/04/file-formats.pdf. [Accessed 03 05 2020].

"Description about the .obj file format," [Online]. Available: http://paulbourke.net/dataformats/obj/. [Accessed 03 05 2020].

"Description about Triangle Mesh," [Online]. Available: https://www.cs.cmu.edu/~quake/triangle.html. [Accessed 03 05 2020].

"Example of reading point cloud and converting it into triangle mesh," [Online]. Available: https://pointcloudlibrary.github.io/documentation/documentation/tutorials/greedy_projection.php. [Accessed 03 03 2020].

Published
2020-10-06