cvg/pyceres

Factor graphs with Ceres in Python

What it solves

It provides a way to use the Ceres Solver—a powerful C++ library for solving non-linear least squares problems—directly within Python. It also implements factor graphs specifically for tasks like bundle adjustment and pose graph optimization, which are critical for 3D reconstruction and robotics.

How it works

Pyceres acts as a set of minimal Python bindings for the C++ Ceres Solver. It allows users to define optimization factors (cost functions) either in Python or in C++, with the others being provided by libraries like PyCOLMAP for specific tasks such as reprojection error and epipolar geometry.

Who it’s for

Researchers and developers working on computer vision, 3D reconstruction, and robotics who need the mathematical optimization capabilities of Ceres Solver without leaving the Python environment.

Highlights

  • Minimal Python bindings for the C++ Ceres Solver.
  • Support for factor graph optimization, including bundle adjustment and pose graph optimization.
  • Integration with PyCOLMAP for a variety of specialized cost functions.
  • Cross-platform support for Linux, macOS (Intel and Apple Silicon), and Windows.

Related

  • Project
    ModelEngine-Group/DataMate

    An enterprise-level data processing platform designed to streamline data collection, cleaning, and annotation for model fine-tuning and RAG retrieval.

  • Project
    testdino-hq/playwright-skill

    A collection of 70 structured AI agent skills (Markdown guides) that teach AI coding agents how to write, debug, and scale production-ready Playwright tests.

  • Project
    deepseek-ai/DeepSelect

    A high-performance TopK kernel implementation that provides 2x to 20x speedup over torch.topk for DeepSeek Sparse Attention and LLM sampling.

  • Project
    feyninc/chonkiejs

    A lightweight and efficient TypeScript text chunking library designed for RAG applications, offering multiple strategies including semantic, recursive, and code-aware splitting.