Patch Diffusion: A General Module for Face Manipulation Detection
Authors: Baogen Zhang, Sheng Li, Guorui Feng, Zhenxing Qian, Xinpeng Zhang3243-3251
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We integrate our PD module into four recent face manipulation detection networks, and carry out the experiments on four popular datasets. The results demonstrate that our PD module is able to boost the performance of the existing networks for face manipulation detection. |
| Researcher Affiliation | Academia | Baogen Zhang1, Sheng Li1 , Guorui Feng2, Zhenxing Qian1, Xinpeng Zhang1 1Fudan University 2Shanghai University {bgzhang19, lisheng, zxqian, zhangxinpeng}@fudan.edu.cn, grfeng@shu.edu.cn |
| Pseudocode | No | The paper describes algorithms and formulations through text and equations but does not provide a clearly labeled pseudocode or algorithm block. |
| Open Source Code | Yes | The code is available at https://github.com/starxchina/Patch-Diffusion |
| Open Datasets | Yes | We select Face Forensics++ (Rossler et al. 2019b) as the training dataset. It is a large-scale face manipulation dataset containing 1000 real videos, in which 720 videos are used for training, 140 videos are reserved for validation and 140 videos are for testing. |
| Dataset Splits | Yes | It is a large-scale face manipulation dataset containing 1000 real videos, in which 720 videos are used for training, 140 videos are reserved for validation and 140 videos are for testing. |
| Hardware Specification | No | The paper does not specify the hardware used for experiments (e.g., specific GPU models, CPU, or memory). It only implies the use of computing resources for deep learning. |
| Software Dependencies | No | The paper does not provide specific version numbers for software dependencies such as Python, PyTorch, or TensorFlow. |
| Experiment Setup | Yes | We set α = 2, 15, 17 for a pair of patches whose ground truth are real and background, fake and background, real and fake, respectively. Other hyperparameters are set by λ = 1, ω = 3, l = 1, ξ = 10 8. |