NVRD Dataset Explorer

Novel Visual Recognition Dataset — Human vs. Model Ratings on Visual Perturbations

Item Explorer

Data Table

Model vs. Human

Rating × Level

Rating Distribution

Category

Object

Perturbation

Level

Sort by

1 / 800

Original

Base image

Perturbed

Perturbed image

Ratings Comparison

HUMAN RATING DISTRIBUTION

1 (Strongly Disagree)7 (Strongly Agree)

All Ratings (Bar Chart)

Human Rating Distribution for This Item

Category

Object

Perturbation

Object	Category	Perturbation	Level	Human Mean	Gemini	GPT-4o	IDEFICS3	Molmo2	Qwen2-VL

Page 1

Model

Color by

All Models vs. Human (Jittered)

Category

Perturbation

Mean Rating by Perturbation Level

Per-Object Rating Curves

Perturbation

Overall Rating Distribution (Human vs. Models)

Rating Distribution by Category

Rating Distribution by Perturbation Type