Research map / Image generation / GANs
ST-GAN: Spatial Transformer Generative Adversarial Networks for Image Compositing
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use ST-GAN geometric correction instead of the initial warp | Indoor object composites · Indoor objects test set (average) | real classification rate: 24.6 % -> 47.0 % | Helped |
| Use adversarial ST-GAN rather than HomographyNet | Indoor object composites · Indoor objects test set (average) | real classification rate: 37.1 % -> 47.0 % | Helped |
| Use adversarial ST-GAN rather than SDM | Indoor object composites · Indoor objects test set (average) | real classification rate: 25.6 % -> 47.0 % | Helped |
| Add iterative warp updates rather than a non-iterative generator (ablation) | ST-GAN · Indoor objects test set (average) | real classification rate: 31.9 % -> 41.5 % | Helped |
| Increase the iterative warp count from one to two (ablation) | ST-GAN · Indoor objects test set (average) | real classification rate: 31.9 % -> 38.2 % | Helped |
| Increase the iterative warp count from two to four (ablation) | ST-GAN · Indoor objects test set (average) | real classification rate: 38.2 % -> 41.5 % | Helped |
| Add end-to-end fine-tuning after sequential warp training (ablation) | ST-GAN with four warps · Indoor objects test set (average) | real classification rate: 41.5 % -> 47.0 % | Helped |
Closest papers by meaning
- Spatial Fusion GAN for Image Synthesis (CVPR 2019 · similarity 0.90)
- A Geometric Analysis of Deep Generative Image Models and Its Applications (ICLR 2021 · similarity 0.88)
- Toward Realistic Image Compositing With Adversarial Learning (CVPR 2019 · similarity 0.88)
- Generative Adversarial Transformers (ICML 2021 · similarity 0.88)
- WarpGAN: Warping-Guided 3D GAN Inversion with Style-Based Novel View Inpainting (NeurIPS 2025 · similarity 0.86)
- SAN: Inducing Metrizability of GAN with Discriminative Normalized Linear Layer (ICLR 2024 · similarity 0.86)
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.