skills.rest
voxel51
voxel51OFFICIAL
Explore all skills by this author →
WebsiteGitHub

fiftyone-embeddings-visualization

Reduce dataset embeddings to 2D or 3D with UMAP, t-SNE, or PCA for visualization in the FiftyOne app.

37|8|Updated Dec 18, 2025
Search Tags:#embeddings#UMAP#PCA#visualize#FiftyOne#t-SNE#dataset-analysis
One-click install
npx skills add https://github.com/voxel51/fiftyone-skills --skill fiftyone-embeddings-visualization
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: fiftyone-embeddings-visualization
Source: https://github.com/voxel51/fiftyone-skills/tree/main/skills/fiftyone-embeddings-visualization
Command: npx skills add https://github.com/voxel51/fiftyone-skills --skill fiftyone-embeddings-visualization

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Visualize high-dimensional dataset embeddings in 2D or 3D to reveal structure and clusters.

Core Features & Use Cases

  • Embedding Visualization: Reduce embeddings to 2D/3D using UMAP, t-SNE, or PCA for interactive exploration.
  • Exploratory Data Analysis: Compare classes, inspect cluster separation, and identify outliers in embedding space.
  • Use Case: For example, visualize CIFAR-like dataset embeddings to observe class separation and detect mislabeled samples.

Quick Start

Use this Skill to visualize embeddings for your dataset:

  • Set context: set_context(dataset_name="my-dataset")
  • Launch the FiftyOne App: launch_app() and wait 5-10 seconds
  • Ensure brain operators are available:
    • list_operators(builtin_only=False)
    • get_operator_schema(operator_uri="@voxel51/brain/compute_visualization")
  • Compute embeddings if not present:
    • execute_operator( operator_uri="@voxel51/brain/compute_similarity", params={ "brain_key": "my_viz", "model": "clip-vit-base32-torch", "embeddings": "clip_embeddings", "backend": "sklearn", "metric": "cosine" } )
  • Generate 2D visualization:
    • execute_operator( operator_uri="@voxel51/brain/compute_visualization", params={ "brain_key": "my_viz", "embeddings": "clip_embeddings", "method": "umap", "num_dims": 2 } )
  • Open the App at http://localhost:5151/ and explore the Embeddings panel with brain key "my_viz".

Frequently Asked Questions about fiftyone-embeddings-visualization

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I visualize high-dimensional dataset embeddings in 2D?▼

You can visualize dataset embeddings by applying a dimensionality reduction method like UMAP, t-SNE, or PCA to project high-dimensional vectors into 2D or 3D space, then exploring the results interactively in the FiftyOne app.

What is the best way to identify clusters and outliers in dataset embeddings?▼

Visualizing dataset embeddings in 2D or 3D using UMAP, t-SNE, or PCA reveals structural clusters and outliers. This exploratory analysis helps you inspect class separation and detect mislabeled samples within the embedding space.

Can I use UMAP or t-SNE for embedding visualization in FiftyOne?▼

Yes, FiftyOne supports UMAP, t-SNE, and PCA for embedding visualization. You execute the compute_visualization operator with your chosen method and num_dims parameter, then explore the reduced embeddings in the app's Embeddings panel.

Do I need to compute embeddings before visualizing them in 2D?▼

Yes, you must compute embeddings before visualization. Use the compute_similarity brain operator with a model like CLIP to generate embeddings, then apply a reduction method such as UMAP, t-SNE, or PCA to visualize them in 2D or 3D.

How do I launch the FiftyOne app to explore reduced embeddings?▼

To explore reduced embeddings, call launch_app() and wait 5-10 seconds for initialization. Open http://localhost:5151/ and select your brain key in the Embeddings panel to interactively visualize dataset clusters and outliers.

Related Skills

Explore More Data & Analytics Skills→
zangjeicyzangjeicy

clip

Classify images and retrieve cross-modal matches using CLIP embeddings.

Community
Intermediate
voxel51voxel51
37

fiftyone-dataset-export

Export FiftyOne datasets to COCO, YOLO, VOC, CVAT, and CSV formats.

Official
Intermediate
voxel51voxel51
37

fiftyone-find-duplicates

Finds and deduplicates near-identical images in FiftyOne datasets using similarity scoring and thresholds.

Official
Advanced
yf8578yf8578
1

umap-learn

Reduce high-dimensional datasets to low-dimensional embeddings using UMAP.

Community
Intermediate