MLflow + FiftyOne Integration

Welcome to the demo repoistory of an example workflow of MLflow integrated into the FiftyOne App!

The demo consists of two main parts:

1. The mlflow_fiftyone_workflow.ipynb notebook. It will guide you through an example workflow of training with YOLOv9 using FiftyOne, Ultralytics, and MLflow!

2. The FiftyOne + MLflow plugin, a FiftyOne plugin that brings MLflow UI in the app as a panel, as well as track experiments and runs across your FiftyOne datasets as seen below!

Helpful Links

Here are some helpful links to resources or tools used within the notebook.

Plugins

• Find image quality issues
• Find exact and approximate duplicates
• Find outliers in your dataset
• Compare one model to another

Find all the other FiftyOne Plugins here!

FiftyOne Brain

• Visualize embeddings with FiftyOne Brain
• Search your datasets with text prompts or sort by similarity
• Sort by uniqueness

YOLOv9

• Original Yolov9 Repo