Research map / Detection and segmentation / CNN features
InstaFormer: Instance-Aware Image-to-Image Translation With Transformer
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 |
|---|---|---|---|
| Use InstaFormer instead of MGUIT | Instance-aware image-to-image translation · INIT (average) | CIS: 1.112 -> 1.149 | Helped |
| Use InstaFormer instead of MGUIT | Instance-aware image-to-image translation · INIT (average) | IS: 1.254 -> 1.353 | Helped |
| Use InstaFormer instead of MGUIT | Instance-aware image-to-image translation · INIT, two-direction average | SSIM: 0.842 -> 0.845 | Helped |
| Use InstaFormer instead of CUT | Instance-aware image-to-image translation · INIT, two-direction average | FID: 78.00 -> 79.05 | Hurt |
| Use InstaFormer instead of MGUIT | Domain-adaptive object detection · KITTI→Cityscapes | mAP: 54.6 -> 55.5 | Helped |
| Use 6 Transformer blocks instead of 3 (ablation) | InstaFormer · INIT sunny→night | FID: 89.96 -> 84.72 | Helped |
| Use 6 Transformer blocks instead of 3 (ablation) | InstaFormer · INIT sunny→night | SSIM: 0.711 -> 0.872 | Helped |
| Use 9 Transformer blocks instead of 6 (ablation) | InstaFormer · INIT sunny→night | FID: 84.72 -> 85.28 | Hurt |
| Use 9 Transformer blocks instead of 6 (ablation) | InstaFormer · INIT sunny→night | SSIM: 0.872 -> 0.879 | Helped |
| Use 8 attention heads instead of 4 (ablation) | InstaFormer · INIT sunny→night | FID: 84.72 -> 81.92 | Helped |
| Use 8 attention heads instead of 4 (ablation) | InstaFormer · INIT sunny→night | SSIM: 0.872 -> 0.873 | Helped |
| Use 4 attention heads instead of 1 (ablation) | InstaFormer · INIT sunny→night | FID: 89.17 -> 84.72 | Helped |
Closest papers by meaning
- Towards Instance-Level Image-To-Image Translation (CVPR 2019 · similarity 0.89)
- InstaGAN: Instance-aware Image-to-Image Translation (ICLR 2019 · similarity 0.88)
- Memory-Guided Unsupervised Image-to-Image Translation (CVPR 2021 · similarity 0.86)
- Masked and Adaptive Transformer for Exemplar Based Image Translation (CVPR 2023 · similarity 0.86)
- AttnLRP: Attention-Aware Layer-Wise Relevance Propagation for Transformers (ICML 2024 · similarity 0.85)
- LaTtE-Flow: Layerwise Timestep-Expert Flow-based Transformer (ICML 2026 · similarity 0.84)
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.