FiftyOne¶
View on GitHub
Join us on Discord
Try it in ColabThe open-source tool for building high-quality datasets and computer vision models
Nothing hinders the success of machine learning systems more than poor qualitydata. And without the right tools, improving a model can be time-consuming andinefficient.
FiftyOne supercharges your machine learning workflows by enabling you tovisualize datasets and interpret models faster and more effectively.
Improving data quality and understanding your model’s failure modes are themost impactful ways to boost the performance of your model.
FiftyOne provides the building blocks for optimizing your dataset analysispipeline. Use it to get hands-on with your data, including visualizing complexlabels, evaluating your models, exploring scenarios of interest, identifyingfailure modes, finding annotation mistakes, and much more!
FiftyOne integrates naturally with your favorite tools. Click on a logo tolearn how:
Note
FiftyOne is growing!Sign up for the mailing listto learn about new features as they come out.
Core Capabilities¶
Curating datasets
Surveys show that machine learning engineers spend over half of their time wrangling data, but it doesn't have to be that way. Use FiftyOne's powerful dataset import and manipulation capabilities to manage your data with ease.
Learn how to load data into FiftyOne
Evaluating models
Aggregate metrics alone don’t give you the full picture of your ML models. In practice, the limiting factor on your model’s performance is often data quality issues that you need to see to address. FiftyOne makes it easy to do just that.
See how to evaluate models with FiftyOne
Visualizing embeddings
Are you using embeddings to analyze your data and models? Use FiftyOne's embeddings visualization capabilities to reveal hidden structure in you data, mine hard samples, pre-annotate data, recommend new samples for annotation, and more.
Experience the power of embeddings
Working with geolocation
Many datasets have location metadata, but visualizing location-based datasets has traditionally required closed source or cloud-based tools. FiftyOne provides native support for storing, visualizing, and querying datasets by location.
Finding annotation mistakes
Annotations mistakes create an artificial ceiling on the performance of your model. However, finding these mistakes by hand is not feasible! Use FiftyOne to automatically identify possible label mistakes in your datasets.
Check out the label mistakes tutorial
Removing redundant images
During model training, the best results will be seen when training on unique data. Use FiftyOne to automatically remove duplicate or near-duplicate images from your datasets and curate diverse training datasets from your raw data.
Try the image uniqueness tutorial
Core Concepts¶
FiftyOne Library¶
FiftyOne’s core library provides a structured yet dynamic representation toexplore your datasets. You can efficiently query and manipulate your dataset byadding custom tags, model predictions and more.
1 2 3 4 5 6 7 8 91011121314 | importfiftyoneasfodataset=fo.Dataset("my_dataset")sample=fo.Sample(filepath="/path/to/image.png")sample.tags.append("train")sample["custom_field"]=51dataset.add_sample(sample)view=dataset.match_tags("train").sort_by("custom_field").limit(10)forsampleinview:print(sample) |
Note
FiftyOne is designed to be lightweight and flexible, making it easy to loadyour datasets. FiftyOne supports loading datasets in a variety of commonformats out-of-the-box, and it also provides the extensibility to loaddatasets in custom formats.
Check outloading datasets to seehow to load your data into FiftyOne.
FiftyOne App¶
The FiftyOne App is a graphical user interface that makes it easy to exploreand rapidly gain intuition into your datasets. You can visualize labels likebounding boxes and segmentations overlaid on the samples; sort, query andslice your dataset into any subset of interest; and more.
FiftyOne Brain¶
The FiftyOne Brain is a library of powerful machine learning-poweredcapabilities that provide insights into your datasets and recommend ways tomodify your datasets that will lead to measurably better performance of yourmodels.
1234 | importfiftyone.brainasfobfob.compute_uniqueness(dataset)rank_view=dataset.sort_by("uniqueness") |
FiftyOne Plugins¶
FiftyOne provides a powerful plugin framework that allows for extending andcustomizing the functionality of the tool to suit your specific needs.
With plugins, you can add new functionality to the FiftyOne App, createintegrations with other tools and APIs, render custom panels, and add custombuttons to menus.
WithFiftyOne Teams, you can even writeplugins that allow users to execute long-running tasks from within the App thatrun on a connected compute cluster.
Dataset Zoo¶
The FiftyOne Dataset Zoo provides a powerful interface for downloading datasetsand loading them into FiftyOne.
It provides native access to dozens of popular benchmark datasets, and it alssupports downloading arbitrary public or private datasets whosedownload/preparation methods are provided via GitHub repositories or URLs.
123456 | importfiftyoneasfoimportfiftyone.zooasfozdataset=foz.load_zoo_dataset("coco-2017",split="validation")session=fo.launch_app(dataset) |
Model Zoo¶
The FiftyOne Model Zoo provides a powerful interface for downloading models andapplying them to your FiftyOne datasets.
It provides native access to hundreds of pre-trained models, and it alsosupports downloading arbitrary public or private models whose definitions areprovided via GitHub repositories or URLs.
1 2 3 4 5 6 7 8 910111213141516171819 | importfiftyoneasfoimportfiftyone.zooasfozdataset=foz.load_zoo_dataset("coco-2017",split="validation",max_samples=50,shuffle=True,)model=foz.load_zoo_model("clip-vit-base32-torch",text_prompt="A photo of a",classes=["person","dog","cat","bird","car","tree","chair"],)dataset.apply_model(model,label_field="zero_shot_predictions")session=fo.launch_app(dataset) |
What’s Next?¶
Where should you go from here? You could…
Try one of thetutorials that demonstrate the uniquecapabilities of FiftyOne
Explorerecipes for integrating FiftyOne intoyour current ML workflows
Check out thecheat sheets for topics you maywant to master quickly
Consult theuser guide for detailed instructions onhow to accomplish various tasks with FiftyOne
Need Support?¶
If you run into any issues with FiftyOne or have any burning questions, feelfree toconnect with us on Discord or reach out tous atsupport@voxel51.com.