FD-DB: Frequency-Decoupled Dual-Branch Network for Unpaired Synthetic-to-Real Domain Translation
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910018473820160 |
|---|---|
| author | Zang, Chuanhai Hu, Jiabao Song, XW |
| author_facet | Zang, Chuanhai Hu, Jiabao Song, XW |
| contents | Synthetic data provide low-cost, accurately annotated samples for geometry-sensitive vision tasks, but appearance and imaging differences between synthetic and real domains cause severe domain shift and degrade downstream performance. Unpaired synthetic-to-real translation can reduce this gap without paired supervision, yet existing methods often face a trade-off between photorealism and structural stability: unconstrained generation may introduce deformation or spurious textures, while overly rigid constraints limit adaptation to real-domain statistics. We propose FD-DB, a frequency-decoupled dual-branch model that separates appearance transfer into low-frequency interpretable editing and high-frequency residual compensation. The interpretable branch predicts physically meaningful editing parameters (white balance, exposure, contrast, saturation, blur, and grain) to build a stable low-frequency appearance base with strong content preservation. The free branch complements fine details through residual generation, and a gated fusion mechanism combines the two branches under explicit frequency constraints to limit low-frequency drift. We further adopt a two-stage training schedule that first stabilizes the editing branch and then releases the residual branch to improve optimization stability. Experiments on the YCB-V dataset show that FD-DB improves real-domain appearance consistency and significantly boosts downstream semantic segmentation performance while preserving geometric and semantic structures. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_09476 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | FD-DB: Frequency-Decoupled Dual-Branch Network for Unpaired Synthetic-to-Real Domain Translation Zang, Chuanhai Hu, Jiabao Song, XW Computer Vision and Pattern Recognition Synthetic data provide low-cost, accurately annotated samples for geometry-sensitive vision tasks, but appearance and imaging differences between synthetic and real domains cause severe domain shift and degrade downstream performance. Unpaired synthetic-to-real translation can reduce this gap without paired supervision, yet existing methods often face a trade-off between photorealism and structural stability: unconstrained generation may introduce deformation or spurious textures, while overly rigid constraints limit adaptation to real-domain statistics. We propose FD-DB, a frequency-decoupled dual-branch model that separates appearance transfer into low-frequency interpretable editing and high-frequency residual compensation. The interpretable branch predicts physically meaningful editing parameters (white balance, exposure, contrast, saturation, blur, and grain) to build a stable low-frequency appearance base with strong content preservation. The free branch complements fine details through residual generation, and a gated fusion mechanism combines the two branches under explicit frequency constraints to limit low-frequency drift. We further adopt a two-stage training schedule that first stabilizes the editing branch and then releases the residual branch to improve optimization stability. Experiments on the YCB-V dataset show that FD-DB improves real-domain appearance consistency and significantly boosts downstream semantic segmentation performance while preserving geometric and semantic structures. |
| title | FD-DB: Frequency-Decoupled Dual-Branch Network for Unpaired Synthetic-to-Real Domain Translation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2602.09476 |