RECORD
Anime-to-Real Clothing: Cosplay Costume Generation via Image-to-Image Translation
Computer Vision and Pattern Recognition · Submitted 26 Aug 2020
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim Experiments demonstrated that, with two types of evaluation metrics, the proposed GAN achieves better performance than existing methods.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
Cosplay has grown from its origins at fan conventions into a billion-dollar global dress phenomenon. To facilitate imagination and reinterpretation from animated images to real garments, this paper presents an automatic costume image generation method based on image-to-image translation. Cosplay items can be significantly diverse in their styles and shapes, and conventional methods cannot be directly applied to the wide variation in clothing images that are the focus of this study. To solve this problem, our method starts by collecting and preprocessing web images to prepare a cleaned, paired dataset of the anime and real domains. Then, we present a novel architecture for generative adversarial networks (GANs) to facilitate high-quality cosplay image generation. Our GAN consists of several effective techniques to fill the gap between the two domains and improve both the global and local consistency of generated images. Experiments demonstrated that, with two types of evaluation metrics, the proposed GAN achieves better performance than existing methods. We also showed that the images generated by the proposed method are more realistic than those generated by the conventional methods. Our codes and pretrained model are available on the web.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Cosplay has grown from its origins at fan conventions into a billion-dollar global dress phenomenon. | background 1.00 | 0.04 | 0.02 | asserted 1.00 |
| s1 | To facilitate imagination and reinterpretation from animated images to real garments, this paper presents an automatic costume image generation method based on image-to-image translation. | contribution 1.00 | 0.83 | 0.07 | asserted 0.98 |
| s2 | Cosplay items can be significantly diverse in their styles and shapes, and conventional methods cannot be directly applied to the wide variation in clothing images that are the focus of this study. | gap 0.60 | 0.10 | 0.02 | asserted 1.00 |
| s3 | To solve this problem, our method starts by collecting and preprocessing web images to prepare a cleaned, paired dataset of the anime and real domains. | method 1.00 | 0.93 | 0.02 | asserted 0.94 |
| s4 | Then, we present a novel architecture for generative adversarial networks (GANs) to facilitate high-quality cosplay image generation. | method 0.60 | 0.90 | 0.04 | asserted 0.99 |
| s5 | Our GAN consists of several effective techniques to fill the gap between the two domains and improve both the global and local consistency of generated images. | method 0.95 | 0.71 | 0.06 | asserted 1.00 |
| s6 | Experiments demonstrated that, with two types of evaluation metrics, the proposed GAN achieves better performance than existing methods. | result 1.00 | 0.96 | 0.42 | asserted 1.00 |
| s7 | We also showed that the images generated by the proposed method are more realistic than those generated by the conventional methods. | result 0.99 | 0.95 | 0.39 | asserted 1.00 |
| s8 | Our codes and pretrained model are available on the web. | contribution 0.82 | 0.64 | 0.03 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
Cosplay has grown from its origins at fan conventions into a billion-dollar global dress phenomenon. To facilitate imagination and reinterpretation from animated images to real garments, this paper presents an automatic costume image generation method based on image-to-image translation. Cosplay items can be significantly diverse in their styles and shapes, and conventional methods cannot be directly applied to the wide variation in clothing images that are the focus of this study. To solve this problem, our method starts by collecting and preprocessing web images to prepare a cleaned, paired dataset of the anime and real domains. Then, we present a novel architecture for generative adversarial networks (GANs) to facilitate high-quality cosplay image generation. Our GAN consists of several effective techniques to fill the gap between the two domains and improve both the global and local consistency of generated images. Experiments demonstrated that, with two types of evaluation metrics, the proposed GAN achieves better performance than existing methods. We also showed that the images generated by the proposed method are more realistic than those generated by the conventional methods. Our codes and pretrained model are available on the web.
Human review
How the literature uses it
1 arXiv paper cite this in 1 sentence
What it cites
Uses the method or tool of 5 papers
-
In (b), we added the coarse-to-fine scheme, where we changed the size of the discriminator's feature maps from $70×70$ to $1×1$.[Our training at (b) (i.e., adding the coarse-to-fine scheme only) became unstable, and hence, we changed the discriminator's instance normalization [ref] to the equalized learning rate [ref] to stabilize the training of (b).
Progressive Growing of GANs for Improved Quality, Stability, and Variation · Comparisons among different configurations with the proposed method · 0.99
-
We applied spectral normalization [ref] to each discriminator to satisfy the Lipschitz constraint.
Spectral Normalization for Generative Adversarial Networks · Improved domain discriminator · 1.00
-
Here, to maintain local consistency, we propose a multi-scale patch discriminator similar to the structure reported in [ref] .
InGAN: Capturing and Remapping the "DNA" of a Natural Image · Real/fake discriminator · 0.90
Cites as background 6 papers
-
To improve the quality of generated images, various GAN techniques have been proposed, such as spectral normalization [ref] , coarse-to-fine schemes [ref] , and useful tips for training [ref] .
Progressive Growing of GANs for Improved Quality, Stability, and Variation · Related work · 0.95
-
To improve the quality of generated images, various GAN techniques have been proposed, such as spectral normalization [ref] , coarse-to-fine schemes [ref] , and useful tips for training [ref] .
Spectral Normalization for Generative Adversarial Networks · Related work · 0.95
-
Collecting images paired across source and target domains is usually time-consuming; hence, unpaired image-to-image translation frameworks have also been studied [ref] .
Attention-Guided Generative Adversarial Networks for Unsupervised Image-to-Image Translation · Related work · 0.92
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.