margelo/react-native-vision-camera

📸 A powerful, high-performance React Native Camera library.

What it solves

VisionCamera provides a high-performance camera interface for React Native applications, allowing developers to integrate advanced camera functionality without sacrificing speed or quality.

How it works

It acts as a powerful wrapper for mobile camera hardware, offering deep control over resolutions, FPS, and device selection. A key feature is its "Frame Processors," which use JS worklets to allow real-time processing of camera frames for tasks like AI object detection and facial recognition.

Who it’s for

Mobile app developers using React Native who need professional-grade camera capabilities, such as high-resolution capture, real-time AI vision tasks, or custom GPU-accelerated filters.

Highlights

  • Real-time AI Integration: Supports Frame Processors for facial recognition and object detection.
  • High Performance: Includes a custom C++/GPU accelerated resizer using Metal and Vulkan.
  • Advanced Capture: Supports 4k/8k images, HDR, Night modes, and customizable FPS up to 240.
  • Extensibility: Allows drawing shapes, text, and shaders directly onto the camera feed.

Related

  • Project
    eduardolundgren/tracking.js

    A lightweight JavaScript library that brings computer vision algorithms like color tracking and face detection directly into the browser.

  • Project
    realsenseai/librealsense

    RealSense SDK 2.0 (librealsense) – open‑source C++ library for Intel RealSense depth cameras, offering cross‑platform colour‑depth streaming, calibration data, and wrappers (Python, ROS 2, C#, Unity). Provides pre‑built binaries or source builds, supports AMD GPU acceleration, and includes sample code for easy start‑up.

  • Project
    roboflow/notebooks

    roboflow/notebooks is a GitHub repository containing 61 Jupyter notebook tutorials for computer vision, covering models like YOLO, SAM, Gemini, and Qwen2.5-VL. Each notebook can be opened in Colab, Kaggle, or SageMaker Studio Lab, and includes links to related blog posts, videos, and papers. It's designed for hands-on learning of tasks like object detection, segmentation, OCR, and data extraction.

  • Project
    roboflow/rf-detr

    RF‑DETR is an open‑source, transformer‑based library for real‑time object detection, instance segmentation, and keypoint detection. Built on a DINOv2 backbone, it offers a family of models (from tiny “Atto” to large “2XL”) that achieve state‑of‑the‑art COCO accuracy while running in 1–17 ms on an NVIDIA T4. Install via `pip install rfdetr`, use a single Python API for all three tasks, and optionally run NAS on the Roboflow platform to tailor a model to your own dataset.