Robustness Meets Deep Learning: An End-to-End Hybrid Pipeline for Unsupervised Learning of Egomotion.
Alex Zihao ZhuWenxin LiuZiyun WangVijay KumarKostas DaniilidisPublished in: CoRR (2018)
Keyphrases
- end to end
- deep learning
- unsupervised learning
- ego motion
- field of view
- deep architectures
- optical flow
- supervised learning
- stereo camera
- image sequences
- camera motion
- semi supervised
- machine learning
- d scene
- deep belief networks
- multi camera
- expectation maximization
- robust estimation
- flow field
- dimensionality reduction
- feature selection
- mental models
- reinforcement learning
- text classification
- computer vision
- real time
- dynamic scenes
- motion field
- multiscale
- vision system
- object recognition
- pattern recognition