Research map / Image generation / Text-to-image diffusion
LoFGAN: Fusing Local Representations for Few-Shot Image Generation
Zheng Gu, Wenbin Li, Jing Huo, Lei Wang, Yang Gao
Read the paperCheck 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 |
|---|---|---|---|
| replace global fusion with the local fusion module (LFM) (ablation) | MatchingGAN† · Flowers FID | FID: 139.90 -> 86.59 | Improved |
| replace global fusion with the local fusion module (LFM) (ablation) | MatchingGAN† · Animal Faces FID | FID: 147.95 -> 112.99 | Improved |
| replace global fusion with the local fusion module (LFM) (ablation) | MatchingGAN† · VGGFace FID | FID: 27.93 -> 22.99 | Improved |
| replace global fusion with the local fusion module (LFM) (ablation) | MatchingGAN† · Flowers LPIPS | LPIPS: 0.3410 -> 0.3704 | Improved |
| add local reconstruction loss to LFM (ablation) | MatchingGAN†+LFM · Flowers FID | FID: 86.59 -> 79.33 | Improved |
| add local reconstruction loss to LFM (ablation) | MatchingGAN†+LFM · VGGFace FID | FID: 22.99 -> 20.31 | Improved |
| add local reconstruction loss to LFM (ablation) | MatchingGAN†+LFM · Animal Faces FID | FID: 112.99 -> 112.81 | No clear effect |
| add local reconstruction loss to LFM (ablation) | MatchingGAN†+LFM · Animal Faces LPIPS | LPIPS: 0.5024 -> 0.4964 | Hurt |
| augment classifier training data with LoFGAN images | ResNet18 classifier · Flowers low-data classification | top-1 accuracy: 60.00 -> 65.10 | Improved |
| augment classifier training data with MatchingGAN† | ResNet18 classifier · VGGFace low-data classification | top-1 accuracy: 67.38 -> 65.17 | Hurt |
| add LFM to MatchingGAN† augmentation (ablation) | ResNet18 classifier · Flowers low-data classification | top-1 accuracy: 60.39 -> 62.75 | Improved |
| increase training shots k_train from 3 to 9 | LoFGAN · Flowers | average FID: 78.30 -> 77.18 | Improved |
Later papers that compared with this method
1 later accepted paper reports LoFGAN as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Where Is My Spot? Few-Shot Image Generation via Latent Subspace Optimization (CVPR 2023 · better on 2 of 2 shared results)
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
- LFS-GAN: Lifelong Few-Shot Image Generation (ICCV 2023 · similarity 0.89)
- Local and Global GANs With Semantic-Aware Upsampling for Image Generation (TPAMI 2022 · similarity 0.88)
- WeditGAN: Few-Shot Image Generation via Latent Space Relocation (AAAI 2024 · similarity 0.88)
- Few-Shot Unsupervised Image-to-Image Translation (ICCV 2019 · similarity 0.88)
- MetaGAN: An Adversarial Approach to Few-Shot Learning (NeurIPS 2018 · similarity 0.88)
- Exact Fusion via Feature Distribution Matching for Few-shot Image Generation (CVPR 2024 · similarity 0.87)
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.
