# Figure sources

Figures are cropped from *Learning to Falsify: Hierarchical Multiple Instance Learning for High Resolution AI-Generated Image Detection*, Haoran Chen, Daiqi Gao, Xinyuan Li, Yongwei Wang, Lei Zhu, Rui Cai, and Jianfeng Dong (ACM Multimedia 2026).

Source: https://arxiv.org/abs/2611.04529 · https://doi.org/10.1145/3767308.3836365

The paper states a **Creative Commons Attribution 4.0 International** license. These crops retain the original figure content; surrounding page text and captions were omitted.

| File | Source |
| --- | --- |
| `motivation.png` | Figure 1, PDF page 2 |
| `method.png` | Figure 2, PDF page 4 |
| `chameleon.png` | Figure 3, PDF page 7 |
| `fatas.png` | Figure 4, PDF page 7 |
| `robustness.png` | Figure 5, PDF page 8 |

Rendered at 216 dpi. The software license is separate from the paper and figure license.