Research map / Detection and segmentation / Open-set and few-shot detection
Towards Instance-Level Image-To-Image Translation
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 |
|---|---|---|---|
| add instance-level object styles and reconstruction to global-style translation | INIT · INIT test set, sunny→night/rainy/cloudy (average) | LPIPS diversity: 0.247 -> 0.274 | Helped |
| add instance-level object styles and reconstruction to global-style translation | INIT · INIT test set, translation directions in table (average) | CIS: 1.032 -> 1.043 | Helped |
| add instance-level object styles and reconstruction to global-style translation | INIT · INIT test set, translation directions in table (average) | IS: 1.166 -> 1.179 | Helped |
| use shared rather than separate discriminators for global images and objects (ablation) | INIT · INIT test set, sunny→night/rainy/cloudy (average) | LPIPS diversity: 0.246 -> 0.274 | Helped |
| use shared rather than separate discriminators for global images and objects (ablation) | INIT · INIT test set, translation directions in table (average) | CIS: 1.039 -> 1.043 | Helped |
| use shared rather than separate discriminators for global images and objects (ablation) | INIT · INIT test set, translation directions in table (average) | IS: 1.167 -> 1.179 | Helped |
| add INIT-generated synthetic images to real COCO training data | Mask R-CNN · MS COCO 2017 val, real validation images | object detection AP IoU 0.5:0.95: +1.1 | Helped |
| add INIT-generated synthetic images to real COCO training data | Mask R-CNN · MS COCO 2017 val, synthetic validation images | object detection AP IoU 0.5:0.95: +2.1 | Helped |
| augment COCO training with INIT-generated rather than MUNIT-generated images | COCO object detection · COCO 2017 detection | IoU: +0.7 -> +1.1 | Helped |
| train and test on synthetic rather than real images | PSPNet with ResNet-50 · Cityscapes scene parsing | mIoU: 76.6\% -> 74.6\% | Hurt |
| add synthetic training data to real data | Mask R-CNN · MS COCO 2017 val | object detection AP: 30.4 -> 32.6 | Helped |
Closest papers by meaning
- DUNIT: Detection-Based Unsupervised Image-to-Image Translation (CVPR 2020 · similarity 0.92)
- Memory-Guided Unsupervised Image-to-Image Translation (CVPR 2021 · similarity 0.90)
- The Spatially-Correlative Loss for Various Image Translation Tasks (CVPR 2021 · similarity 0.90)
- InstaFormer: Instance-Aware Image-to-Image Translation With Transformer (CVPR 2022 · similarity 0.89)
- SHUNIT: Style Harmonization for Unpaired Image-to-Image Translation (AAAI 2023 · similarity 0.88)
- Domain Adaptive Image-to-Image Translation (CVPR 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.