Skip to content

Dataset · google

Google's DreamBooth benchmark dataset for object generation

google/dreambooth

Photos of 30 diverse objects and pets from multiple angles for personalizing text-to-image models.

What's inside

The official dataset from Google's DreamBooth paper, containing image-text pairs (fewer than 1,000 records). It features 30 unique subjects across 15 classes: 9 live subjects (dogs and cats) and 21 objects, with 4–6 photos per subject captured under different conditions, environments, and angles.

What you can do with it

  • Testing personalization methods and LoRA training for diffusion models
  • Comparing subject identity preservation in image generation
Size
up to 1K rows
Downloads this month
4.9K
Likes
57

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("google/dreambooth")