3DB: A Framework for Debugging Computer Vision Models
Guillaume Leclerc, Hadi Salman, Andrew Ilyas, Sai Vemprala, Logan Engstrom, Vibhav Vineet, Kai Yuanqing Xiao, Pengchuan Zhang, Shibani Santurkar, Greg Yang, Ashish Kapoor, Aleksander Madry
摘要
We introduce 3DB: an extendable, uni ed framework for testing and debugging vision models using photorealistic simulation. We demonstrate, through a wide range of use cases, that 3DB allows users to discover vulnerabilities in computer vision systems and gain insights into how models make decisions. 3DB captures and generalizes many robustness analyses from prior work, and enables one to study their interplay. Finally, we nd that the insights generated by the system transfer to the physical world. We are releasing 3DB as a library 1 alongside a set of example analyses 2 , guides 3 , and documentation 4 . * Work partially completed while at Microsoft Research. † Equal contribution.
问问这篇 Paper
智能体会读完全文。
Lune 把这篇 Paper 索引到了每一个公式,引用它的顶会 Paper 也一样。你提问,回答直接引用原文。
引用它的顶会 Paper15
- Salient ImageNet: How to discover spurious features in Deep Learning?Sahil Singla, Soheil FeiziICLR 2022 · 被引用 144 次
- 3D Common Corruptions and Data AugmentationOguzhan Fatih Kar, Teresa Yeo, Andrei Atanov, Amir ZamirCVPR 2022 · 被引用 80 次
- Datamodels: Understanding Predictions with Data and Data with PredictionsAndrew Ilyas, Sung Min Park, Logan Engstrom, Guillaume Leclerc 等ICML 2022 · 被引用 66 次
- Adaptive Testing of Computer Vision ModelsIrena Gao, Gabriel Ilharco, Scott M. Lundberg, Marco Túlio RibeiroICCV 2023 · 被引用 49 次
- Missingness Bias in Model DebuggingSaachi Jain, Hadi Salman, Eric Wong, Pengchuan Zhang 等ICLR 2022 · 被引用 45 次
它引用的顶会 Paper12
- Habitat: A Platform for Embodied AI ResearchManolis Savva, Jitendra Malik, Devi Parikh, Dhruv Batra 等ICCV 2019 · 被引用 1,863 次
- Hypersim: A Photorealistic Synthetic Dataset for Holistic Indoor Scene UnderstandingMike Roberts, Jason Ramapuram, Anurag Ranjan, Atulit Kumar 等ICCV 2021 · 被引用 633 次
- Noise or Signal: The Role of Image Backgrounds in Object RecognitionKai Yuanqing Xiao, Logan Engstrom, Andrew Ilyas, Aleksander MadryICLR 2021 · 被引用 451 次
- On Completeness-aware Concept-Based Explanations in Deep Neural NetworksChih-Kuan Yeh, Been Kim, Sercan Ömer Arik, Chun-Liang Li 等NeurIPS 2020 · 被引用 390 次
- Meta-Sim: Learning to Generate Synthetic DatasetsAmlan Kar, Aayush Prakash, Ming-Yu Liu, Eric Cameracci 等ICCV 2019 · 被引用 272 次
相关 Paper
- Unadversarial Examples: Designing Objects for Robust VisionHadi Salman, Andrew Ilyas, Logan Engstrom, Sai Vemprala 等NeurIPS 2021 · 被引用 65 次
- RoboMD: Uncovering Robot Vulnerabilities through Semantic Potential FieldsSom Sagar, Jiafei Duan, Sreevishakh Vasudevan, Yifan Zhou 等ICLR 2026 · 被引用 7 次
- Universal 3-Dimensional Perturbations for Black-Box Attacks on Video Recognition SystemsShangyu Xie, Han Wang, Yu Kong, Yuan HongS&P 2022 · 被引用 32 次
- LIBERO-Plus: A Progressive Robustness Benchmark for Visual-Language-Action ModelsSenyu Fei, Siyin Wang, Junhao Shi, Zihao Dai 等CVPR 2026
- Robo3D: Towards Robust and Reliable 3D Perception against CorruptionsLingdong Kong, Youquan Liu, Xin Li, Runnan Chen 等ICCV 2023 · 被引用 151 次
