Research map / Image generation / GANs
Reusing Discriminators for Encoding: Towards Unsupervised Image-to-Image Translation
Runfa Chen, Wenbing Huang, Binghui Huang, Fuchun Sun, Bin Fang
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| reuse discriminator features as the target-domain encoder (ablation) | NICE-GAN · cat ↔ dog | FID: 80.75 -> 63.80; 64.36 -> 51.49 | Improved |
| add residual attention to discriminator features (ablation) | NICE-GAN · dog ↔ cat | FID: 80.75 -> 67.60 | Improved |
| add residual attention to discriminator features (ablation) | NICE-GAN · cat → dog | FID: 51.49 -> 44.67 | Hurt |
| decrease shared encoder-discriminator layers by 1 (ablation) | NICE-GAN · dog → cat | FID: 48.79 -> 48.55 | Improved |
| increase shared encoder-discriminator layers by 1 (ablation) | NICE-GAN · dog → cat | FID: 48.79 -> 53.52 | Hurt |
| decrease shared encoder-discriminator layers by 1 (ablation) | NICE-GAN · cat → dog | FID: 44.67 -> 52.92 | Hurt |
| increase shared encoder-discriminator layers by 1 (ablation) | NICE-GAN · cat → dog | FID: 44.67 -> 54.90 | Hurt |
| remove global-scale classifier C² (ablation) | NICE-GAN · dog → cat | FID: 48.79 -> 203.56 | Hurt |
| remove global-scale classifier C² (ablation) | NICE-GAN · cat → dog | FID: 44.67 -> 238.62 | Hurt |
| remove middle- and global-scale classifiers C¹ and C² (ablation) | NICE-GAN · dog → cat | FID: 48.79 -> 216.03 | Hurt |
| remove middle- and global-scale classifiers C¹ and C² (ablation) | NICE-GAN · cat → dog | FID: 44.67 -> 231.24 | Hurt |
| use NICE-GAN instead of the strongest reported FID baseline | unsupervised image translation · zebra → horse | FID: 145.47 -> 149.48 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DRIT — DRIT++: Diverse Image-to-Image Translation via Disentangled Representations (IJCV 2020) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- TraVeLGAN: Image-To-Image Translation by Transformation Vector Learning (CVPR 2019 · similarity 0.90)
- Discriminative Region Proposal Adversarial Networks for High-Quality Image-to-Image Translation (ECCV 2018 · similarity 0.90)
- Discriminative Region Proposal Adversarial Network for High-Quality Image-to-Image Translation (IJCV 2019 · similarity 0.90)
- Unsupervised Image-to-Image Translation Networks (NeurIPS 2017 · similarity 0.89)
- Rethinking the Paradigm of Content Constraints in Unpaired Image-to-Image Translation (AAAI 2024 · similarity 0.89)
- Compressing Image-to-Image Translation GANs Using Local Density Structures on Their Learned Manifold (AAAI 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
