Research map / Image generation / Editing and style transfer
Multimodal Structure-Consistent Image-to-Image Translation
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 |
|---|---|---|---|
| train detector on structure-consistent multimodal AugGAN data vs MUNIT | YOLO · GTA night detection | AP: 22.8 -> 33.0 | Helped |
| Multimodal AugGAN vs best baseline | Faster R-CNN · SYNTHIA night detection | AP: 72.6 -> 73.5 | Helped |
| Multimodal AugGAN vs best baseline | YOLO · BDD100k-val-night B+G | AP: 39.3 -> 41.9 | Helped |
| Multimodal AugGAN vs best baseline | Faster R-CNN · BDD100k-val-night B+G | AP: 64.4 -> 67.0 | Helped |
| Multimodal AugGAN FCN training data vs CyCADA | FCN8s · SYNTHIA night segmentation | mIoU: 60.1 -> 62.6 | Helped |
| use multimodal GAN (AugCGAN) vs unimodal baselines | YOLO · SYNTHIA night detection | AP: 39.5 -> 28.2 | Hurt |
| multimodal only vs unimodal only training data (ablation) | YOLO/Faster R-CNN · BDD100k-val-night | AP: YOLO 37.1 -> 38.9; FRCN 61.2 -> 60.0 | Mixed |
| mix unimodal and multimodal data vs unimodal only (ablation) | YOLO/Faster R-CNN · BDD100k-val-night | AP: YOLO 37.1 -> 37.9; FRCN 61.2 -> 62.0 | Helped |
| fine-tune unimodal-pretrained detector on multimodal data vs unimodal only (ablation) | YOLO · BDD100k-val-night | AP: 37.1 -> 40.4 | Helped |
| fine-tune unimodal-pretrained detector on multimodal data vs unimodal only (ablation) | Faster R-CNN · BDD100k-val-night | AP: 61.2 -> 64.5 | Helped |
| diversity vs MUNIT | Multimodal AugGAN · BDD100k-val-day | LPIPS: 0.31 ± 0.108 -> 0.47 ± 0.155 | Helped |
| generated night images vs real night images | YOLO · BDD100k-val-night | AP: 40.4 vs 40.5 for 2k real | No clear effect |
Closest papers by meaning
- Latent Filter Scaling for Multimodal Unsupervised Image-To-Image Translation (CVPR 2019 · similarity 0.89)
- Harmonizing Maximum Likelihood with GANs for Multimodal Conditional Generation (ICLR 2019 · similarity 0.88)
- Breaking the Cycle - Colleagues Are All You Need (CVPR 2020 · similarity 0.88)
- TraVeLGAN: Image-To-Image Translation by Transformation Vector Learning (CVPR 2019 · similarity 0.88)
- Unsupervised Image-to-Image Translation With Generative Prior (CVPR 2022 · similarity 0.87)
- Rethinking the Paradigm of Content Constraints in Unpaired Image-to-Image Translation (AAAI 2024 · similarity 0.85)
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.