Research map / Label-efficient and robust learning / Domain adaptation and generalization
Bridging Day and Night: Target-Class Hallucination Suppression in Unpaired Image Translation
Shuwei Li, Lei Tan, Robby T. Tan
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 |
|---|---|---|---|
| add hallucination segmentation loss and suppression loss (ablation) | Ours · BDD100K day-to-night (Faster R-CNN) | mAP: 14.11 -> 17.40 | Improved |
| add suppression loss Lsupp (given Lhl) (ablation) | Ours · BDD100K day-to-night (Faster R-CNN) | mAP: 16.43 -> 17.40 | Improved |
| add hallucination loss Lhl (given Lsupp) (ablation) | Ours · BDD100K day-to-night (Faster R-CNN) | mAP: 15.55 -> 17.40 | Improved |
| add both losses for traffic light class (ablation) | Ours · BDD100K day-to-night (Faster R-CNN) traffic light | AP: 4.48 -> 8.44 | Improved |
| translated data from generic unpaired I2I (DDIB, InstructPix2Pix) vs daytime-only detector | DDIB/InstructPix2Pix · BDD100K day-to-night (Faster R-CNN) | mAP: 13.75 -> 10.67 | Hurt |
| hallucination-suppressed translation vs best instance-aware baseline | Ours · BDD100K day-to-night (Faster R-CNN) | mAP: 15.08 -> 17.40 | Improved |
| hallucination-suppressed translation vs InstaFormer | Ours · KITTI to Cityscapes | mAP: 55.5 -> 57.4 | Improved |
| hallucination-suppressed translation vs InstaFormer on car | Ours · KITTI to Cityscapes car | AP: 69.5 -> 69.5 | No clear effect |
Closest papers by meaning
- Hallucination-aware Intermediate Representation Editing in Large Vision-Lanugage Models (ICLR 2026 · similarity 0.91)
- One Token, Two Fates: A Unified Framework via Vision Token Manipulation Against MLLMs Hallucination (CVPR 2026 · similarity 0.91)
- Reducing Hallucinations in Large Vision-Language Models via Latent Space Steering (ICLR 2025 · similarity 0.91)
- Analyzing and Mitigating Object Hallucination: A Training Bias Perspective (AAAI 2026 · similarity 0.91)
- Beyond the Global Scores: Fine-Grained Token Grounding as a Robust Detector of LVLM Hallucinations (CVPR 2026 · similarity 0.90)
- Image Token Matters: Mitigating Hallucination in Discrete Tokenizer-based Large Vision-Language Models via Latent Editing (NeurIPS 2025 · similarity 0.90)
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.
