Research map / Adversarial robustness and security / Attacks on text and graphs
Attacking Video Recognition Models with Bullet-Screen Comments
Kai Chen, Zhipeng Wei, Jingjing Chen, Zuxuan Wu, Yu-Gang Jiang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use RL instead of basin hopping to select BSC positions and transparency (ablation) | BSC attack · UCF-101, C3D | FR (%): 65.8 -> 90.1 | Improved |
| Use RL instead of random BSC position and transparency selection (ablation) | BSC attack · UCF-101, C3D | FR (%): 68.8 -> 90.1 | Improved |
| Use RL instead of basin hopping to select BSC positions and transparency (ablation) | BSC attack · UCF-101, LRCN | FR (%): 97.4 -> 99.5 | Improved |
| Use RL instead of random BSC position and transparency selection (ablation) | BSC attack · UCF-101, I3D-Slow | FR (%): 89.6 -> 96.5 | Improved |
| Use RL instead of basin hopping to select BSC positions and transparency (ablation) | BSC attack · HMDB-51, C3D | FR (%): 81.4 -> 91.2 | Improved |
| Use RL instead of random BSC position and transparency selection (ablation) | BSC attack · HMDB-51, LRCN | FR (%): 93.9 -> 99.0 | Improved |
| Use BSCs instead of a white square patch against LGS (ablation) | Video attack · UCF-101, C3D | FR (%): 54.2 -> 67.9 | Improved |
| Use BSCs instead of a white square patch against LGS (ablation) | Video attack · HMDB-51, I3D-Slow | FR (%): 67.3 -> 93.9 | Improved |
| Set the number of BSCs to 4 instead of 2 (ablation) | BSC attack on C3D · UCF-101 test videos | FR (%): 68.3 -> 79.2 | Improved |
| Increase the number of BSCs from 4 to 6 (ablation) | BSC attack on C3D · UCF-101 test videos | FR (%): 79.2 -> 73.3 | Hurt |
| Increase font size from 8 to 11 (ablation) | BSC attack on C3D · UCF-101 test videos | FR (%): 79.2 -> 82.1 | Improved |
| Increase font size from 11 to 13 (ablation) | BSC attack on C3D · UCF-101 test videos | FR (%): 82.1 -> 76.2 | Hurt |
Closest papers by meaning
- Sparse Black-Box Video Attack with Reinforcement Learning (IJCV 2022 · similarity 0.89)
- Heuristic Black-Box Adversarial Attacks on Video Recognition Models (AAAI 2020 · similarity 0.88)
- Temporal-Distributed Backdoor Attack against Video Based Action Recognition (AAAI 2024 · similarity 0.86)
- LogoStyleFool: Vitiating Video Recognition Systems via Logo Style Transfer (AAAI 2024 · similarity 0.85)
- Chain of Attack: On the Robustness of Vision-Language Models Against Transfer-Based Adversarial Attacks (CVPR 2025 · similarity 0.85)
- FrameShield: Adversarially Robust Video Anomaly Detection (NeurIPS 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
