Towards Robust Gan-Generated Image Detection: A Multi-View Completion Representation

Authors: Chi Liu, Tianqing Zhu, Sheng Shen, Wanlei Zhou

IJCAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluated the generalization ability of our framework across six popular GANs at different resolutions and its robustness against a broad range of perturbation attacks. The results confirm our method s improved effectiveness, generalization, and robustness over various baselines.
Researcher Affiliation Academia 1School of Computer Science, University of Technology Sydney, Australia 2School of Electrical and Information Engineering, The University of Sydney, Australia 3City University of Macau, Macao SAR, China
Pseudocode No The paper describes the proposed framework and its components but does not include any clearly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not provide a statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets Yes We choose the large-scale facial image dataset Celeb A [Liu et al., 2015] and its high-quality version Celeb A-HQ [Karras et al., 2018] to perform evaluations at different resolutions. ... All the four GANs are pretrained with Celeb A. In the high-resolution setting, we adopt the dataset released by [He et al., 2021] 2, which includes images generated by Pro GAN, Style GAN, and Style GAN2. Note that the Pro GAN and Style GAN are pre-trained with Celeb A-HQ, while the Style GAN2 with another facial image dataset FFHQ [Karras et al., 2019]. (Footnote 2 points to: https://github.com/SSAW14/Beyondthe Spectrum)
Dataset Splits No The paper's Table 1 details 'Training' and 'Test' data sizes but does not explicitly mention a 'validation' set or its specific split percentage/counts for reproduction.
Hardware Specification No The paper does not provide any specific details about the hardware used for running the experiments (e.g., GPU models, CPU types, or memory).
Software Dependencies No The paper mentions software components like U-Net, Xception, and Adam optimizer, but does not provide specific version numbers for these or other relevant libraries/frameworks (e.g., PyTorch, TensorFlow).
Experiment Setup Yes We train the whole framework with a batch size of 80 using the Adam optimizer [Kingma and Ba, 2015]. The initial learning rate is 1e-3, and we reduce it to half after every ten epochs. τ in Eq. 8, and λ in Eq. 2 are empirically set to 4 and 10, respectively. We also use random Gaussian noise, color jitter, and blurring for data augmentation on the restorer side.