Semantic Image Synthesis with Unconditional Generator
Authors: JungWoo Chae, Hyunin Cho, Sooyeon Go, Kyungmook Choi, Youngjung Uh
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments validate advantages of our method on a range of datasets: human faces, animal faces, and buildings. Code will be available online soon. We demonstrate pixel-level content creation with a pretrained generator on several datasets including Celeb AMask-HQ, LSUN Church, and LSUN Bedroom. We also present real image editing examples where the source image is edited to conform to the shape of the target image or target mask. Lastly, our method achieves noticeably higher m IOU scores and significantly lower FID than prior works. Furthermore, our method outperforms the baselines in both qualitative and quantitative measures. |
| Researcher Affiliation | Collaboration | Jung Woo Chae12 Hyunin Cho1 Sooyeon Go1 Kyungmook Choi1 Youngjung Uh1 1Yonsei Unviersity, Seoul, South Korea 2LG CNS AI Research, Seoul, South Korea |
| Pseudocode | No | The paper describes the model architecture and training process textually and with mathematical formulas, but does not include any explicit pseudocode or algorithm blocks. |
| Open Source Code | No | Code will be available online soon. |
| Open Datasets | Yes | We have conducted a range of experiments to highlight the adaptability of our approach across diverse datasets, including Celeb AMask-HQ [14], LSUN Church, LSUN Bedroom [41], FFHQ [10], and AFHQ [3]. |
| Dataset Splits | No | The paper refers to using datasets and conducting experiments, and mentions a "one-shot setting", but does not explicitly provide specific training, validation, and test dataset splits (e.g., percentages or exact counts) or describe a cross-validation setup in the main text. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU models, CPU types, or memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions using Style GAN2 and SPADE architectures but does not specify software dependencies with version numbers (e.g., Python, PyTorch, CUDA versions). |
| Experiment Setup | No | The paper states, "For more detailed information about these experiments and settings, please refer to the Appendix." This indicates that specific experimental setup details like hyperparameters are not present in the main text. |