*** description: >- Overview of image data curation with NeMo Curator including loading, processing, filtering, and export workflows categories: * workflows tags: * image-curation * tar-archives * filtering * embedding * workflows personas: * data-scientist-focused * mle-focused difficulty: beginner content\_type: workflow modality: image-only *** # About Image Curation Learn how to curate high-quality image datasets using NeMo Curator's powerful image processing pipeline. NeMo Curator enables you to efficiently process large-scale image-text datasets, applying quality filtering, content filtering, and semantic deduplication at scale. ## Use Cases * Prepare high-quality image datasets for training generative AI models such as LLMs, VLMs, and WFMs * Curate datasets for text-to-image model training and fine-tuning * Process large-scale image collections for multimodal foundation model pretraining * Apply quality control and content filtering to remove inappropriate or low-quality images * Generate embeddings and semantic features for image search and retrieval applications * Remove duplicate images from large datasets using semantic deduplication ## Architecture NeMo Curator's image curation follows a modular pipeline architecture where data flows through configurable stages. Each stage performs a specific operation and passes processed data to the next stage in the pipeline. ```mermaid flowchart LR A[Tar Archive Input] --> B[File Partitioning] B --> C[Image Reader
DALI GPU-accelerated] C --> D[CLIP Embeddings
ViT-L/14] D --> E[Aesthetic Filtering
Quality scoring] E --> F[NSFW Filtering
Content filtering] F --> G[Duplicate Removal
Semantic deduplication] G --> H[Export & Sharding
Tar + Parquet output] classDef input fill:#e1f5fe,stroke:#0277bd,color:#000 classDef processing fill:#f3e5f5,stroke:#7b1fa2,color:#000 classDef output fill:#e8f5e8,stroke:#2e7d32,color:#000 class A input class B,C,D,E,F,G processing class H output ``` This pipeline architecture provides: * **Modularity**: Add, remove, or reorder stages based on your workflow needs * **Scalability**: Distributed processing across multiple GPUs and nodes using Ray * **Flexibility**: Configure parameters for each stage independently * **Efficiency**: GPU-accelerated processing with DALI and CLIP models ## Introduction Master the fundamentals of NeMo Curator's image curation pipeline and set up your processing environment. Learn about ImageBatch, ImageObject, and pipeline stages for efficient image curation data-structures distributed architecture Learn prerequisites, setup instructions, and initial configuration for image curation setup configuration quickstart ## Curation Tasks ### Load Data Load and process large-scale image datasets from local storage using tar archives with GPU-accelerated DALI for efficient distributed processing. Load and process JPEG images from tar archives using DALI tar-archives dali gpu-accelerated ### Process Data Transform and enhance your image data through embeddings, classification, and filters. Generate image embeddings using CLIP models. embeddings Apply built-in filters for aesthetic quality and NSFW content filtering. Aesthetic NSFW quality filtering Remove duplicate images using semantic similarity and clustering. deduplication semantic clustering ### Pipeline Management Optimize and manage your image curation pipelines with advanced execution backends and resource management. Configure Ray-based executors for distributed processing and resource management. ray distributed resource-management Optimize performance with DALI GPU acceleration and efficient resource allocation. dali gpu-acceleration performance ### Save & Export Export your curated image datasets with metadata preservation, custom resharding options, and support for downstream training pipelines. Save metadata to Parquet and export filtered datasets with custom resharding. parquet tar-archives resharding