Despite the popular use of object association in many cell tracking problems, there has not been an open implementation of this method, let alone a well-designed library with high extensibility and standard interfaces. We attempt to address this situation by introducing BOT, a extensible C++ library for biomedical object tracking. Though initially designed for cell tracking, we hope to make BOT a generic algorithmic library adaptable to various biomedical tracking applications. Briefly, BOT features: 1. Solver - An efficient pairwise object association solver based on integer linear programming (ILP); 2. Feature - A rich set of generic features and extensible feature design using factory pattern; 3. Learning - Structured learning for high-dimensional feature parametrization; 4. Diversity - Confiurable workflow for supporting diverse applications; 5. GUI - A user-friendly GUI by integration with the Interactive Learning and Segmentation Toolkit (ilastik, http://www.ilastik.org/). This library is the accompanying code for our paper: Xinghua Lou, Fred A. Hamprecht. Structured Learning for Cell Tracking. In Twenty-Fifth Annual Conference on Neural Information Processing Systems, 2011. (NIPS'11) To find the paper as well as the design document and user guide, please visit http://hci.iwr.uni-heidelberg.de/Staff/xlou/research/tracking.html or, take a look at ./doc. Note: This library has been compiled and tested under 1. Linux 32-bit and 64-bit; 2. Windows 32-bit; 3. Mac (by Jonathan Mackenzie, [email protected]) If you find any problem compiling or running it under other platforms, please let the author know. We thank the following users for their useful feedback: * Jan Funke for spotting errors in the TrckingTrainer class; * Bernhard Kausler for many suggestins; * Jonathan Mackenzie for bug reports; * Sébastien Tosi for help on incorporating libtiff. (C) Xinghua Lou ([email protected], [email protected])