dhryou/VHA-bench
VHA-bench VHA-bench is a human-annotated hallucination benchmark for generative image super-resolution. It contains 90 RealSR reference images and outputs from StableSR, DiffBIR, SeeSR, OSEDiff, PURE, and VOSR: 630 images forming 540 GT–SR pairs. The human study contains 11,200 individual ratings from 200 participants. Scores range from 1 to 5; higher values indicate more hallucination relative to the ground truth. Paper Code Project page VHA+ checkpoint Load image… See the full description on the dataset page: https://huggingface.co/datasets/dhryou/VHA-bench.
| Author / organization | dhryou |
|---|---|
| Downloads | 0 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 21 October 2026 |
| Last updated at source | 21 October 2026 |
Connected in the knowledge graph
- dhryou/VHA-plusModel · used to train
- VHA: VQA-based Hallucination Assessment for Generative Image Super-ResolutionPaper · cites
Recent changes
- 21 Oct 2026 Description: VHA+ Lightweight hallucination assessment for generative image super-resolution. Paper: https://arxiv.org/abs/2610.25770 Code: https://github.com/dhryougit/Hallucination-Assessment → VHA-bench VHA-bench is a human-annotated hallucination benchmark for generative image super-resolution. It contains 90 RealSR reference images and outputs from StableSR, DiffBIR, SeeSR, OSEDiff, PURE, and VOSR: 630 images forming 540 GT–SR
- 21 Oct 2026 Topics added task_categories:image-to-image, super-resolution, hallucination-assessment, image-quality-assessment
Changes detected by Silytics between refreshes of the source.
Related papers
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 21 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.