LibreYOLO/libreyolo

LibreYOLO is a MIT licensed open source computer vision library

What it solves

LibreYOLO provides a unified, MIT-licensed interface for a vast array of computer vision tasks. It eliminates the need to learn different APIs for different models, allowing users to perform detection, segmentation, pose estimation, and more using a single, consistent codebase.

How it works

The library wraps numerous vision models (such as YOLOv9, SAM, and Depth Anything) behind a single API. Users can switch between different vision tasks simply by changing the model checkpoint file. It supports various input sources, including static images, webcams, RTSP streams, and YouTube URLs, and provides built-in tools for training and exporting models to multiple deployment formats.

Who it’s for

It is designed for computer vision developers and researchers who need a versatile tool for rapid prototyping, training, and deploying a wide variety of vision models without managing multiple disparate libraries.

Highlights

  • Unified API: One consistent interface for 17 different vision tasks, including OCR, depth estimation, and background removal.
  • Extensive Model Support: Includes a massive library of models across detection, segmentation, classification, and VLMs (Vision Language Models).
  • Flexible Deployment: Supports exporting to 12 different formats, including ONNX, TensorRT, and CoreML, and integrates with NVIDIA Triton.
  • Comprehensive Training: Built-in support for multi-GPU training, LoRA, distillation, and integration with popular logging tools like Weights & Biases and MLflow.

Related

  • Project
    ultralytics/ultralytics

    A comprehensive computer vision framework featuring YOLO models for real-time object detection, segmentation, classification, and pose estimation.

  • Project
    Geekgineer/YOLOs-CPP

    A production-ready C++ inference engine that provides a unified API for the entire YOLO family, supporting tasks from object detection to metric depth estimation.

  • Project
    Koldim2001/YOLO-Patch-Based-Inference

    A Python library that enables patch-based inference for YOLO models to improve the detection of small objects in large images through image tiling and result consolidation.

  • Project
    roboflow/roboflow-python

    The official Python package for Roboflow, enabling developers to programmatically manage computer vision datasets, train models, and run inference.

  • Project
    ultralytics/yolov5

    YOLOv5 is a PyTorch‑based, open‑source object detection/segmentation/classification model suite from Ultralytics, offering multiple model sizes, simple Python/CLI inference, one‑line training, and export to many deployment formats.