Research map / Image generation / Autoregressive and latent generation
TherA: Thermal-Aware Visual-Language Prompting for Controllable RGB-to-Thermal Infrared Translation
Dong-Guw Lee, Tai Hyoung Rhee, Hyunsoo Jang, Young-Sik Shin, Ukcheol Shin, Ayoung Kim
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 |
|---|---|---|---|
| condition diffusion on TherA-VLM thermal embedding instead of its text schema (ablation) | TherA · M3FD | PSNR: 13.23 -> 18.24 | Improved |
| condition diffusion on TherA-VLM thermal embedding instead of its text schema (ablation) | TherA · M3FD | FID: 142.37 -> 105.52 | Improved |
| condition on thermal embedding instead of generic LLaVA (ablation) | TherA · FLIR | FID: 101.82 -> 112.93 | Hurt |
| condition on thermal embedding instead of text schema (ablation) | TherA · FLIR | PSNR: 14.65 -> 16.56 | Improved |
| use structured canonical schema text rather than generic VLM (ablation) | TherA · M3FD | PSNR: 11.85 -> 13.23 | Improved |
| replace InstructPix2Pix with TherA-VLM-conditioned latent diffusion (ablation) | InstructPix2Pix · M3FD (fine-tuned) | PSNR: 13.94 -> 19.54 | Improved |
| train on R2T2 and evaluate zero-shot vs other translators | TherA · M3FD (zero-shot) | PSNR: 12.84 -> 18.24 | Improved |
| use structured segmentation and scene-description priors | DiffV2IR · M3FD (fine-tuned) | PSNR: 18.97 vs 17.36 (F-ViTA) | Improved |
| set image guidance scale sV=0.5 and VLM guidance sS=1.5 (ablation) | TherA · M3FD | translation quality | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- InstructPix2Pix — InstructPix2Pix: Learning To Follow Image Editing Instructions (CVPR 2023) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ThermalGen: Style-Disentangled Flow-Based Generative Models for RGB-to-Thermal Image Translation (NeurIPS 2025 · similarity 0.89)
- CoVisIT: Cross-modal Prior Guided Diffusion Model for Visible-to-Infrared Image Translation (NeurIPS 2026 · similarity 0.89)
- D3F-IR: Dual-Domain Deterministic Flow Matching for Visible-to-Infrared Translation (ECCV 2026 · similarity 0.89)
- Thermo-JEPA: Learning a Geometry-Grounded Thermal World Model via Cross-Modal Privileged Masking (ECCV 2026 · similarity 0.88)
- Enhancing Infrared Vision: Progressive Prompt Fusion Network and Benchmark (NeurIPS 2025 · similarity 0.87)
- ThermalGaussian: Thermal 3D Gaussian Splatting (ICLR 2025 · similarity 0.86)
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.
