MangaGAN: Unpaired Photo-to-Manga Translation Based on The Methodology of Manga Drawing
Authors: Hao Su, Jianwei Niu, Xuefeng Liu, Qingfeng Li, Jiahe Cui, Ji Wan2611-2619
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments show that Manga GAN can produce high-quality manga faces preserving both the facial similarity and manga style, and outperforms other reference methods. |
| Researcher Affiliation | Academia | 1State Key Lab of VR Technology and System, School of Computer Science and Engineering, Beihang University 2Industrial Technology Research Institute, School of Information Engineering, Zhengzhou University 3Hangzhou Innovation Institute, Beihang University |
| Pseudocode | No | The paper describes methods through text and diagrams, but does not include any explicit pseudocode or algorithm blocks. |
| Open Source Code | No | The paper states: 'The Mange GAN-BL dataset is available for academic use.' which refers to the dataset, but does not include a statement about open-source code availability or a link to a code repository for the methodology itself. |
| Open Datasets | Yes | Dm, named Manga GANBL, is a new dataset we collected from a popular manga work Bleach... The Mange GAN-BL dataset is available for academic use. Dp contains 1197 front view of face photos collected from CFD (Ma, Correll, and Wittenbrink 2015). |
| Dataset Splits | No | The paper describes the datasets used (Dm, Dp, Db) and mentions 'each training data is converted to grayscale', but does not provide specific percentages, sample counts, or methodology for training/validation/test dataset splits. |
| Hardware Specification | Yes | We implemented Manga GAN in Py Torch (Paszke et al. 2017) and all experiments are performed on a computer with an NVIDIA Tesla V100 GPU. |
| Software Dependencies | No | We implemented Manga GAN in Py Torch (Paszke et al. 2017). |
| Experiment Setup | Yes | For all experiments, we set α1=10, α{2,3}=5, α4=1 in Eq.(5); β{1,3,5}=10, β{2,4,6}=1, and parameters of LSP in Eq.(3) are fixed at λI=1, λpool5=1, λi=0, i {pool1, pool2, pool3, pool4} with the output resolution of 256 256. Moreover, we employ the Adam solver (Kingma and Ba 2014) with a batch size of 5. All networks use the learning rate of 0.0002 for the first 100 epochs, where the rate is linearly decayed to 0 over the next 100 epochs. |