Research map / Image generation / GANs
Learning From Simulated and Unsupervised Images Through Adversarial Training
Check 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 |
|---|---|---|---|
| Refine UnityEyes training images with SimGAN | Gaze CNN · MPIIGaze | mean eye gaze estimation error (degrees): 11.2 -> 7.8 | Helped |
| Refine UnityEyes training images with SimGAN | Gaze CNN · MPIIGaze | % of images within d=7 degrees: 62.3 -> 69.4 | Helped |
| Refine UnityEyes images and increase training data to 4x | Gaze CNN · MPIIGaze | % of images within d=7 degrees: 64.9 -> 87.2 | Helped |
| Refine synthetic training images with SimGAN | Hand pose CNN · NYU hand pose test set | % of images within d=5 pixels: 69.7 -> 72.4 | Helped |
| Use refined synthetic images instead of real labeled training images | Hand pose CNN · NYU hand pose test set | % of images within d=5 pixels: 74.5 -> 72.4 | Hurt |
| Refine synthetic training images and train on 3x data | Hand pose CNN · NYU hand pose test set | % of images within d=5 pixels: 77.7 -> 83.3 | Helped |
| Add a history buffer of refined images to discriminator training (ablation) | SimGAN · MPIIGaze | mean eye gaze estimation error (degrees): 12.2 -> 7.8 | Helped |
| Use local patch adversarial loss instead of global whole-image adversarial loss (ablation) | SimGAN · NYU hand pose images | visual realism | Helped |
| Use mean RGB channels as the self-regularization feature transform instead of the identity map (ablation) | SimGAN color image refinement · Color image refinement | visual realism | Helped |
Closest papers by meaning
- Forward Super-Resolution: How Can GANs Learn Hierarchical Generative Models for Real-World Distributions (ICLR 2023 · similarity 0.88)
- Cluster-Guided Image Synthesis With Unconditional Models (CVPR 2022 · similarity 0.88)
- Unsupervised Pixel-Level Domain Adaptation With Generative Adversarial Networks (CVPR 2017 · similarity 0.88)
- Self-Supervised GANs via Auxiliary Rotation Loss (CVPR 2019 · similarity 0.88)
- Instance-Conditioned GAN (NeurIPS 2021 · similarity 0.88)
- The Unmet Promise of Synthetic Training Images: Using Retrieved Real Images Performs Better (NeurIPS 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.