Self-supervised Adversarial Training of Monocular Depth Estimation against Physical-World Attacks.
Zhiyuan ChengCheng HanJames LiangQifan WangXiangyu ZhangDongfang LiuPublished in: CoRR (2024)
Keyphrases
- physical world
- depth estimation
- monocular images
- depth cues
- stereo vision
- depth map
- virtual world
- ubiquitous computing
- stereo matching
- depth estimates
- depth information
- dynamic scenes
- image sequences
- real scenes
- scene understanding
- super resolution
- pose estimation
- disparity map
- stereo pair
- human beings
- three dimensional
- feature matching
- post processing
- vision system
- high quality