原文传递
Deep Learning Techniques to Estimate 3D Position in Stereoscopic Imagery.
| 题名： | Deep Learning Techniques to Estimate 3D Position in Stereoscopic Imagery. |
| 作者： | Nicholson, J. I. |
| 摘要： | Current automated aerial refueling (AAR) efforts utilize machine vision algorithmsto estimate the pose of a receiver aircraft. However, these algorithms are dependent onseveral conditions such as the availability of precise 3D aircraft models; the accuracyof the pipeline significantly degrades in the absence of high-quality information givenbeforehand. We propose a deep learning architecture that estimates the 3D positionof an object based on stereoscopic imagery. We investigate the use of both machinelearning techniques and neural networks to directly regress the 3D position of thereceiver aircraft. We present a new position estimation framework that is basedon the differences between two stereoscopic images without relying on the stereoblock matching algorithm. We analyze the speed and accuracy of its predictionsand demonstrate the effectiveness of the architecture in mitigating various visualocclusions. |
| 总页数： | 94 pages |
相关文献
- Small-Scale Surf Zone Geometric Roughness.
- Unmanned Aerial Vehicle (UAV) Operators’ Workload Reduction: The Effect of 3D Audio on Operators’ Workload and Performance During Multi-Aircraft Control.
- 4D DRIVE-THROUGH VISUALIZATION OF 1-280 FOR REVIEW OF PROPOSED SIGNING
- 3D Nanostructure Design and Fabrication.
- 3D Navigation and Integrated Hazard Display in Advanced Avionics: Workload, Performance, and Situation Awareness
- 3D Navigation and Integrated Hazard Display in Advanced Avionics: Workload, Performance, and Situation Awareness.
检索历史
应用推荐
