luigifreda/pyslam

pySLAM is a hybrid Python/C++ Visual SLAM pipeline supporting monocular, stereo, and RGB-D cameras. It provides a broad set of modern local and global feature extractors, multiple loop-closure strategies, a volumetric reconstruction module, integrated depth-prediction models, and semantic segmentation capabilities for enhanced scene understanding.

What it solves

pySLAM provides a modular and extensible framework for Visual Simultaneous Localization And Mapping (SLAM). It solves the complexity of prototyping and experimenting with different VO/SLAM techniques by integrating a wide array of local and global features, depth prediction models, and semantic segmentation tools into a single, interoperable Python/C++ environment.

How it works

The project implements a hybrid pipeline that supports monocular, stereo, and RGBD cameras. It features a modular sparse-SLAM core available in both Python (for flexibility) and C++ (for performance), with maps being interoperable between the two. The pipeline integrates:

  • Feature Tracking: A variety of classical and modern local features and matchers.
  • Loop Closing: Methods using descriptor aggregators (BoW, VLAD) and global descriptors (NetVLAD, CosPlace, etc.).
  • Volumetric Reconstruction: A pipeline that converts depth and color images into dense reconstructions using TSDF with voxel hashing or incremental Gaussian Splatting.
  • AI Integration: Built-in support for depth prediction (e.g., DepthAnythingV2) and semantic segmentation (e.g., YOLO, Segformer) to enhance scene understanding.
  • 3D Inference: A feed-forward pipeline for reconstructing 3D scenes from multiple images using models like DUSt3R and Mast3r.

Who it’s for

It is designed for researchers and developers who need a flexible baseline framework to experiment with, prototype, and develop Visual Odometry (VO) and SLAM pipelines.

Highlights

  • Hybrid Implementation: Seamlessly switch between Python and C++ for the sparse-SLAM core.
  • Extensive Model Support: Integrates numerous state-of-the-art depth prediction and semantic segmentation models.
  • Multi-Camera Support: Works with monocular, stereo, and RGBD camera setups.
  • Dense Mapping: Supports volumetric integration and Gaussian Splatting for high-quality 3D reconstructions.
  • Broad Dataset Compatibility: Built-in support for over 10 major dataset types, including KITTI, TUM, and EuRoC.

Related

  • Project
    PeterFWS/Structure-PLP-SLAM

    A visual SLAM system that uses points, lines, and planes for efficient sparse mapping and localization across monocular, RGB-D, and stereo cameras.

  • Project
    stella-cv/stella_vslam

    stella_vslam is an open‑source visual SLAM library (monocular, stereo, RGB‑D) that supports many camera models, lets you save/load maps, and offers modular APIs plus ROS integration. It’s a community fork of OpenVSLAM, actively maintained under a BSD‑2 license.

  • Project
    VSLAM-LAB/VSLAM-LAB

    A comprehensive framework for simplifying the development, evaluation, and benchmarking of Visual SLAM systems across various baselines and datasets.

  • Project
    ACFR-RPG/DynOSAM

    DynoSAM is a Stereo/RGB-D Visual Odometry pipeline for Dynamic SLAM that estimates camera poses and object motions while mapping static backgrounds.

  • Project
    luohongk/SuperVINS

    A real-time Visual-Inertial SLAM framework that replaces traditional handcrafted features with SuperPoint and LightGlue deep learning models to handle challenging imaging conditions.