tracksdata 0.1.0rc2

A Python library for multi-object tracking data structures and algorithms

TracksData

A common data structure and basic tools for multi-object tracking.

Features

• Graph-based representation of tracking problems
• In-memory (RustWorkX) and database-backed (SQL) graph backends
• Nodes and edges can take arbitrary attributes
• SQLGraph backend can index frequently queried attributes for faster filtering
• Standardize API for node operators (e.g. defining objects and their attributes)
• Standardize API for edge operators (e.g. creating edges between nodes)
• Basic tracking solvers: nearest neighbors and integer linear programming
• Compatible with Cell Tracking Challenge (CTC) format
• Efficient subgraphing based on attributes on any graph backend
• Integration with cell tracking evaluation metrics

Installation

pip install tracksdata

Why tracksdata?

TracksData provides a common data structure for multi-object tracking problems. It uses graphs to represent detections (nodes) and their connections (edges), making it easier to work with tracking data across different algorithms.

Key benefits:

• Consistent data representation for tracking problems
• Modular components that can be combined as needed
• Support for both small datasets (in-memory) and large datasets (database)

Documentation

• Full Documentation
• Installation
• Core Concepts
• Getting Started
• API Reference
• FAQ