Get Started with Image Curation#

This guide helps you set up and get started with NeMo Curator’s image curation capabilities. Follow these steps to prepare your environment and run your first image curation pipeline.

Prerequisites#

To use NeMo Curator’s image curation modules, ensure you meet the following requirements:

  • Python 3.10 or higher

    • packaging >= 22.0

  • Ubuntu 22.04/20.04

  • NVIDIA GPU (required for all image modules)

    • Volta™ or higher (compute capability 7.0+)

    • CUDA 12 (or above)

Note

All image curation modules require a GPU. Some text-based modules don’t require a GPU, but image curation does.

Installation Options#

You can install NeMo Curator in three ways:

Install the image modules from PyPI:

pip install nemo-curator[image]

Install the latest version directly from GitHub:

git clone https://github.com/NVIDIA/NeMo-Curator.git
cd NeMo-Curator
pip install "./NeMo-Curator[image]"

NeMo Curator is available as a standalone container:

Warning

Container Availability: The standalone NeMo Curator container is currently in development. Check the NGC Catalog for the latest availability and container path.

# Pull the container 
docker pull nvcr.io/nvidia/nemo-curator:latest

# Run the container
docker run --gpus all -it --rm nvcr.io/nvidia/nemo-curator:latest

See also

For details on container environments and configurations, see Container Environments.

Download Default Configuration#

NeMo Curator provides example pipelines in the Image Curation Tutorial Notebook. You can adapt the code there for your own datasets.

Set Up Data Directory#

Create a directory to store your image datasets:

mkdir -p ~/nemo_curator/data

Basic Image Curation Example#

Here’s a simple example to get started with NeMo Curator’s image curation pipeline:

import nemo_curator as nc
from nemo_curator.datasets import ImageTextPairDataset
from nemo_curator.image.embedders import TimmImageEmbedder
from nemo_curator.image.classifiers import AestheticClassifier

# Path to your dataset (WebDataset format)
dataset_path = "~/nemo_curator/data/mscoco/{00000..00001}.tar"  # Example pattern
id_col = "key"

# Load the dataset
image_dataset = ImageTextPairDataset.from_webdataset(dataset_path, id_col)

# Create image embeddings using a CLIP model
embedding_model = TimmImageEmbedder(
    "vit_large_patch14_clip_quickgelu_224.openai",
    pretrained=True,
    batch_size=1024,
    num_threads_per_worker=16,
    normalize_embeddings=True,
    autocast=False,
)
image_dataset = embedding_model(image_dataset)

# Annotate with aesthetic scores
aesthetic_classifier = AestheticClassifier()
image_dataset = aesthetic_classifier(image_dataset)

# Filter images with aesthetic score > 6
image_dataset.metadata["passes_aesthetic_check"] = image_dataset.metadata["aesthetic_score"] > 6
image_dataset.to_webdataset("~/nemo_curator/data/curated", filter_column="passes_aesthetic_check")

Next Steps#

Explore the Image Curation documentation.