Research map / Image generation / GANs
Generative Image Modeling Using Style and Structure Adversarial Networks
Xiaolong Wang, Abhinav Gupta
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 |
|---|---|---|---|
| add S2-GAN pretraining instead of training from scratch | detector · NYUv2 RGBD object detection (AP, 19 classes) | mean AP (%): 30.9 -> 32.4 | Improved |
| use S2-GAN discriminator instead of DCGAN pretraining | detector · NYUv2 RGBD object detection (AP, 19 classes) | mean AP (%): 30.4 -> 32.4 | Improved |
| use S2-GAN instead of DCGANv2 pretraining | detector · NYUv2 RGBD object detection (AP, 19 classes) | mean AP (%): 31.1 -> 32.4 | Improved |
| use ImageNet pretrained AlexNet instead of S2-GAN pretraining | detector · NYUv2 RGBD object detection (AP, 19 classes) | mean AP (%): 32.4 -> 37.6 | Hurt |
| use S2-GAN features instead of DCGAN discriminator for scene classification | SVM on features · SUN RGB-D scene classification | accuracy (%): +8.2% | Improved |
| generate images with Style and Structure decomposition instead of DCGAN | generator · AMT user study (1000 pairs) | preferred as more realistic (%): 71% | Improved |
| generate images with Style and Structure decomposition instead of DCGAN | generator · Fast-RCNN detections on generated images | average detections per image (0.3 threshold): 1.72 -> 2.2 | Improved |
| generate images with Style and Structure decomposition | generator · Places-AlexNet confidence on generated images | max softmax probability: around 2% | Improved |
Closest papers by meaning
- Analyzing and Improving the Image Quality of StyleGAN (CVPR 2020 · similarity 0.90)
- A Geometrical Perspective on Image Style Transfer with Adversarial Learning (TPAMI 2020 · similarity 0.89)
- LR-GAN: Layered Recursive Generative Adversarial Networks for Image Generation (ICLR 2017 · similarity 0.88)
- Cluster-Guided Image Synthesis With Unconditional Models (CVPR 2022 · similarity 0.88)
- Learning Generative Models of Textured 3D Meshes From Real-World Images (ICCV 2021 · similarity 0.88)
- Synthesizing Coupled 3D Face Modalities by Trunk-Branch Generative Adversarial Networks (ECCV 2020 · 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.
