Joint Unsupervised Learning of Depth, Pose, Ground Normal Vector and Ground Segmentation by a Monocular Camera Sensor
| Autor: | Junqiao Zhao, Wei Tian, Yongkun Wen, Yuyao Huang, Lu Xiong |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
0209 industrial biotechnology
Computer science ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION ground segmentation 02 engineering and technology lcsh:Chemical technology unsupervised learning Biochemistry Article Analytical Chemistry 020901 industrial engineering & automation ego-motion 0202 electrical engineering electronic engineering information engineering Computer vision Segmentation lcsh:TP1-1185 ground normal vector Electrical and Electronic Engineering Instrumentation Monocular business.industry Mutual information Atomic and Molecular Physics and Optics Benchmark (computing) Unsupervised learning RGB color model 020201 artificial intelligence & image processing Artificial intelligence Joint (audio engineering) business unsupervised learning scene depth ego-motion ground segmentation ground normal vector Normal scene depth |
| Zdroj: | Sensors (Basel, Switzerland) Sensors Volume 20 Issue 13 Sensors, Vol 20, Iss 3737, p 3737 (2020) |
| ISSN: | 1424-8220 |
| Popis: | We propose a completely unsupervised approach to simultaneously estimate scene depth, ego-pose, ground segmentation and ground normal vector from only monocular RGB video sequences. In our approach, estimation for different scene structures can mutually benefit each other by the joint optimization. Specifically, we use the mutual information loss to pre-train the ground segmentation network and before adding the corresponding self-learning label obtained by a geometric method. By using the static nature of the ground and its normal vector, the scene depth and ego-motion can be efficiently learned by the self-supervised learning procedure. Extensive experimental results on both Cityscapes and KITTI benchmark demonstrate the significant improvement on the estimation accuracy for both scene depth and ego-pose by our approach. We also achieve an average error of about 3° for estimated ground normal vectors. By deploying our proposed geometric constraints, the IOUaccuracy of unsupervised ground segmentation is increased by 35% on the Cityscapes dataset. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|