README.md

---
pretty_name: 'Pix2Fact: When Vision Is Not Enough — Benchmarking Fine-Grained VQA
  with Web Verification on High-Resolution Real-World Scenes'
dataset_info:
  features:
  - name: image
    dtype: image
  - name: question
    dtype: string
  - name: answer
    dtype: string
  - name: image_url
    dtype: string
  - name: index
    dtype: string
  - name: ItemID
    dtype: string
  - name: is_original_qa
    dtype: string
  - name: if_search_first
    dtype: string
  - name: search_query
    dtype: string
  - name: local_image_path
    dtype: string
  - name: image_description
    dtype: string
  - name: bounding_box
    dtype: string
  - name: evidence_1
    dtype: string
  - name: evidence_2
    dtype: string
  - name: evidence_3
    dtype: string
  - name: evidence_url_1
    dtype: string
  - name: evidence_url_2
    dtype: string
  - name: evidence_url_3
    dtype: string
  - name: caption
    dtype: string
  - name: category
    dtype: string
  - name: confidence
    dtype: string
  - name: rebalanced
    dtype: string
  - name: image_resolution
    dtype: string
  splits:
  - name: train
    num_bytes: 8240548814
    num_examples: 1000
  download_size: 8230119774
  dataset_size: 8240548814
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

# Pix2Fact: When Vision Is Not Enough — Benchmarking Fine-Grained VQA with Web Verification on High-Resolution Real-World Scenes

🌐 **Website:** https://fanfan7589.github.io/pix2fact/

📄 **Paper:** https://arxiv.org/abs/2602.00593

Pix2Fact is a visual question-answering benchmark designed to assess expert-level visual
perception and knowledge search. It comprises **1,000 high-resolution (4K+) images** spanning
eight real-world scenarios, with question–answer pairs meticulously crafted by PhD-holding
annotators. Each question requires both **fine-grained visual grounding** and the integration
of **external (web) knowledge**.

## Usage

```python
from datasets import load_dataset

ds = load_dataset("pix2fact/Pix2FactBenchmark", split="train")
print(ds[0]["question"], ds[0]["answer"])
ds[0]["image"]  # PIL.Image
```

## Fields

- `image` — the high-resolution scene image
- `question` / `answer` — the QA pair
- `image_url` — stable CDN URL of the image (`.../resolve/main/images/<file>`), also downloadable as a file in the repo
- `category` — one of the eight scenario categories
- `search_query`, `evidence_*`, `evidence_url_*` — supporting search queries / evidence
- `image_description`, `caption`, `bounding_box`, `image_resolution`, and other metadata