Illusion of Depth: Revealing Hidden Stereo Vision Vulnerabilities in Depth Estimation
Sri Hrushikesh Varma Bhupathiraju, Tetsu Ishizue, Nicholas U. Costagliola, Ozora Sako, Kentaro Yoshioka, Takeshi Sugawara, Sara Rampazzi
Abstract
As described in the paper, we reveal intrinsic vulnerabilities in the stereo cameras' inherent pixel-sampling and calibration processes, which directly affect the outputs of stereo-matching algorithms. This artifact provides the data and scripts to evaluate the attack on SDE algorithms using the KITTI dataset and high-speed scenarios in the CARLA environment, the stereo point-cloud data collected from RealSense D435 and ZED2 cameras in real-world experiments to validate the manipulated depth, and the scripts to test and verify the proposed defense methodology.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Related papers
- DepthVanish: Optimizing Adversarial Interval Structures for Stereo-Depth-Invisible PatchesYun Xing, Yue Cao, Nhat Chung, Jie M. Zhang et al.NeurIPS 2025
- Cheating Stereo Matching in Full-Scale: Physical Adversarial Attack Against Binocular Depth Estimation in Autonomous DrivingKangqiao Zhao, Shuo Huai, Xurui Song, Jun LuoAAAI 2026
- Invisible but Detected: Physical Adversarial Shadow Attack and Defense on LiDAR Object DetectionRyunosuke Kobayashi, Kazuki Nomoto, Yuna Tanaka, Go Tsuruoka et al.USENIX Security 2025
- DoubleStar: Long-Range Attack Towards Depth Estimation based Obstacle Avoidance in Autonomous SystemsCe Zhou, Qiben Yan, Yan Shi, Lichao SunUSENIX Security 2022
- Weaponizing Reflectivity for Pointcloud Deception with Forged Invisible GeometriesHengwei Chen, Menglan Hu, Tianyue ZhengS&P 2026
