Nucleus
The mission control for your data.
Nucleus
Public Database
My Dataset
camera:FRONTweather:sunny
14 Images SelectedCreate Slice
COCO Dataset
COCO is a large-scale object detection, segmentation, and captioning dataset. It contains 330K images (>200K labeled), 1.5 million object instances, and 80 object categories.
Explore NowOpen Images
Open Images is a dataset of ~9M images annotated with image-level labels, object bounding boxes, object segmentation masks, and visual relationships.
Explore NowBDD100K
The Berkeley DeepDrive dataset contains over 100K videos of driving experiences, each running 30 seconds at 30 frames per second.
Explore Now
Understanding your data = better ML
Aggregate metrics in ML are not good enough. To improve production ML, you need to understand their qualitative failure modes, fix them by gathering more data, and curate diverse scenarios.
Scale Nucleus helps you:
- Visualize your data
- Curate interesting slices within your dataset
- Review and manage annotations
- Measure and debug your model performance
Nucleus is a new way—the right way—to develop ML models, helping us move away from the concept of one dataset and towards a paradigm of collections of scenarios.
Ensure Dataset Health
Understand the health of your datasets by identifying under or overrepresented labels and attributes to mitigate bias without building internal tools.
Automate Data Curation
Choose what data to label automatically, based on model performance and attributes of the raw data. Use visual search to quickly identify interesting edge cases.
Track Model Performance
Easily upload models via API. Share data queries across teams, track performance history, compare models, and debug model performance for faster development.
How It Works
Nucleus provides advanced tooling for understanding, visualizing, curating, and collaborating on your data – allowing teams to build better ML models via a powerful interface and APIs.
- Upload your data
- Slice your data
- Upload Model Predictions
- Compare Performance
- Model Debug
- Sorting by IOU
Nucleus
I trained EfficientNet on PandaSet, check out the results! It does pretty well for cars but not for cyclists.
This is great! Let's share with the team and train it on images with more cyclists to improve that performance
CollaborationBeta
Scale Nucleus enables collaboration across teams between ML engineers, PMs, operations, and managers. By sharing whole datasets, dataset slices, model predictions, and more, Nucleus is a single source of truth for data within the team, increasing iteration speed.
Api
Robust APIs enable you to integrate your entire machine learning development cycle with Scale Nucleus.
Explore Our Docs1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16client.uploadModelPredictions({ annotations: [ { label: 'car', type: 'box', x: 655.2530151367188, y: 269.39984741210935, width: 70.560546875, height: 50.55670166015625, reference_id: 'put anything you like here', confidence: 0.7009941479671303, } ], }, (err, cb) => { // put something here })
Scale AI’s mission is to accelerate the development of AI applications.
Trusted by leading machine learning teams to deliver high-quality training data, Scale Nucleus provides the tools for ML teams to manage their entire ML lifecycle from dataset management, data selection, data annotation, to model development - all in one place.
Nucleus currently supports Image data, with support for 3D Sensor Fusion, Video, Text and Document data on the roadmap.
Get Access To Nucleus Today
