Integrations

Use FiftyOne to download, visualize, and evaluate ActivityNet dataset, a large-scale video benchmark for human activity understanding.

Use Albumentations transformation pipelines with FiftyOne to apply and visualize augmentations, and test their effectiveness on your data.

Deploy FiftyOne in your enterprise via Docker or Kubernetes.

Download, visualize, and evaluate the COCO dataset, a large-scale object detection, segmentation, and captioning dataset, directly in FiftyOne.

CVAT is a popular open-source image and video annotation tool. Upload data directly from FiftyOne to CVAT to add or edit labels.

Work with datasets residing in the public cloud (e.g. AWS, Google Cloud, Microsoft Azure …) or your own private cloud.

Use FiftyOne datasets to train a model with Detectron2, Facebook AI Research’s library for detection and segmentation algorithms.

Elasticsearch’s vector database offers an efficient way to create, store, and search vector embeddings at scale. Create Elasticsearch indexes, upload vectors, and run similarity queries programmatically and via the FiftyOne App.

Run inference with Hugging Face Transformers models on your FiftyOne datasets, as well as directly push and load datasets from Hugging Face Hub.

Label Studio is a data labeling tool supporting images, video, text, and other modalities. Upload data directly from FiftyOne to Label Studio for labeling.

Labelbox is a cloud-based image and video annotation tool. Upload data directly from FiftyOne to Labelbox for labeling.

LanceDB is a serverless vector database with deep integrations with the Python ecosystem. Create LanceDB tables and run similarity queries programmatically or via the FiftyOne App.

Train Flash models on FiftyOne datasets and improve your models

Milvus makes unstructured data search more accessible and provides a consistent user experience. Create Milvus collections, upload vectors, and run similarity queries programmatically and via the FiftyOne App.

MongoDB
Atlas Vector Search enables building semantic search and AI-powered applications. Create MongoDB vector search indexes, add/remove vectors, and run similarity queries programmatically and via the FiftyOne App.

FiftyOne supports web-based or cloud interactive development environments such as Jupyter Notebook, Google Colab, Databricks Notebook, SageMaker Notebook, and more.

Use FiftyOne, the recommended integration by Google, to download and visualize their Open Images dataset.

Run inference with CLIP models on FiftyOne datasets

Pinecone provides a fully managed, developer-friendly, easily scalable vector database. Create Pinecone indexes, upload vectors, and run similarity queries programmatically and via the FiftyOne App.

Load any model from PyTorch hub and run inference on FiftyOne datasets

Qdrant vector database & vector similarity search engine provides search for the nearest high-dimensional vectors. Create Qdrant collections, upload vectors, and run similarity queries, both programmatically and via the FiftyOne App.

Redis Vector Search enables the storage of vectors and the associated metadata within hashes or JSON to perform vector similarity searches. Create Redis vector search indexes, upload vectors, and run similarity queries programmatically or via the FiftyOne App.

FiftyOne Teams-specific methods for managing users, dataset permissions, plugins, API keys, and more.

Segments.ai is a data labeling platform providing 2D, 3D point cloud, and multi-sensor labeling. Request Segments.ai annotations directly from FiftyOne.

Run inference with YOLO-NAS models on FiftyOne datasets.

Load, fine-tune, and run inference with Ultralytics models on FiftyOne datasets.

V7 is a popular image and video annotation tool. Upload images or video data directly from FiftyOne to V7 to add or edit labels.

Track model experiments with Weights & Biases and visualize model predictions in FiftyOne to find the best-performing model.