metadata
task_categories:
- image-to-text
dataset_info:
features:
- name: id
dtype: string
- name: image_1
dtype: image
- name: image_2
dtype: image
- name: choices
struct:
- name: A
dtype: string
- name: B
dtype: string
- name: C
dtype: string
- name: D
dtype: string
- name: ground_truth
dtype: string
- name: category
dtype: string
splits:
- name: train
num_bytes: 4188405701.612
num_examples: 1756
download_size: 5002312163
dataset_size: 4188405701.612
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
VDiff-Bench
VDiff-Bench is a multiple-choice benchmark for fine-grained visual difference identification. Each example presents two similar images and four candidate descriptions, exactly one of which states a real difference between the images.
Dataset structure
The train split contains 1,756 questions with the following fields:
id: stable example identifier.image_1,image_2: the paired images.choices: an object containing answer choicesA,B,C, andD.ground_truth: label of the correct choice.category: visual change category.
The public release intentionally excludes source identifiers, intermediate difference annotations, generated-negative metadata, and other dataset-construction fields.
Change categories
| Category | Questions |
|---|---|
| Change of position | 158 |
| Change of motion | 138 |
| Change of color (regional) | 170 |
| Change of color (whole image) | 147 |
| Appear / disappear | 201 |
| Change of noise / resolution | 150 |
| Change of texture | 177 |
| Substitution and size | 199 |
| OCR | 221 |
| Illumination | 195 |
Loading
from datasets import load_dataset
dataset = load_dataset("elaine1wan/image_diff_data", split="train")
example = dataset[0]
image_1 and image_2 are decoded as PIL images by default. Benchmark performance is measured using exact-match accuracy on ground_truth.
Citation
Citation information will be added with the accompanying paper release.