FiftyOne Labs

FiftyOne Labs brings research solutions and experimental features for machine learning.

Table of Features

This repository contains a curated collection of FiftyOne Labs Features which are developed using the FiftyOne plugins ecosystem. These features are organized into the following categories:

• Machine Learning Lab: core machine learning experimental features
• Visualization Lab: features for advanced visualization

Machine Learning Lab

Name	Tags	Description
@51labs/labs_panel	ml utils	A panel listing all the available FiftyOne Labs features
@51labs/video_apply_model	ml video	Apply image model to video dataset using torch dataloader
@51labs/few_shot_learning	ml classification	Interactive few-shot learning with multiple model types
@51labs/label_propagation	Propagating Labels across frames of a video
@51labs/box_combine	ml detection	Weighted Boxes Fusion for detections
@51labs/zero-shot-coreset-selection	ml	Zero-shot coreset selection (ZCore) for unlabeled image data
@51labs/click_segmentation	ml segmentation	Interactive image segmentation via prompts

Visualization Lab

Name	Description
@51labs/viz_placeholder	Placeholder for visualization feature

Using FiftyOne Labs

Install FiftyOne

If you haven't already, install FiftyOne:

pip install fiftyone

Installing specific FiftyOne Labs Feature

To install all the features in this repository, you can run:

fiftyone labs install --all

You can also install specific FiftyOne Labs features using:

fiftyone labs install <name1> <name2> ...

Installing via Labs Panel

Labs Panel offers a convenient interface to install FiftyOne Labs features in the FiftyOne App. To get started, install the Labs Panel:

fiftyone labs install @51labs/labs_panel

FiftyOne Labs CLI

For more command line tools for FiftyOne Labs, check out the CLI documentation.

Feedback

For questions, comments, and suggestions, head to the fiftyone-labs Discord Channel.

Contributing

Check out the contributions guide for more information.