Intel RealSense SDK 2.0

repository·master·Indexed 27 days ago

https://github.com/realsenseai/librealsense

A cross-platform library for Intel RealSense depth cameras. It provides C and C++ APIs, as well as wrappers for Python (pyrealsense2), C#, Unity, and ROS 2. The SDK includes high-level pipeline APIs for streaming, processing blocks for depth enhancement (spatial, temporal, and decimation filters), and tools for recording and playback using ROS-bag files. It supports various platforms including Ubuntu, Windows, macOS, NVIDIA Jetson, and Raspberry Pi.

Tokens
80.5K
Snippets
191
Records
448
Agent score
93%

What's inside librealsense

  1. Overview of GLFW

    master

    GLFW is an open-source, multi-platform library designed for developing applications using OpenGL, OpenGL ES, and Vulkan. It provides a platform-independent API for tasks such as creating windows and contexts, surfaces, reading input, and handling events.

    Supported platforms include:

    • Windows
    • macOS
    • Linux and other Unix-like systems
    • Experimental support for the Wayland protocol.
  2. Overview of UE4 Runtime Mesh Component (RMC)

    master

    The RuntimeMeshComponent (RMC) is a specialized Unreal Engine 4 component designed for rendering and collision on meshes generated at runtime. It is intended for use cases such as voxel engines, custom model viewers, or modding support.

    Compared to the standard ProceduralMeshComponent or CustomMeshComponent, RMC offers:

    • Efficiency: Uses approximately 1/3 the memory and provides faster mesh data updates (capable of updating 600k+ vertices in real time).
    • Features: Supports up to 8 UV channels, high precision normals, tessellation, navigation mesh support, and fully configurable vertex structures.
    • Collision: Supports both static collision (e.g., terrain) and dynamic collision, with full support for async collision cooking.
    • Conversion: Supports RMC <-> StaticMeshComponent conversions (SMC -> RMC at runtime/editor; RMC -> SMC in editor).
    • Optimization: Includes a static render path for meshes that update infrequently and allows visibility/shadowing configuration per section.
  3. Overview of D400 Series RealSense features

    master

    The D400 series RealSense cameras are designed for robotics and drones, providing hardware-accelerated depth sensing. Key features include:

    • Hardware Rectification: Devices come fully calibrated, producing hardware-rectified image pairs.
    • High Frame Rate: Depth calculations are performed on-camera at up to 90 FPS.
    • Accuracy: Supports sub-pixel accuracy and high fill-rate.
    • Texture Projector: Includes an on-board texture projector to improve depth accuracy in low-texture or dark environments.
    • Low Power: Operates on standard USB 5V power, drawing approximately 1-1.5 W.
  4. Overview of RS400 Advanced Mode

    master

    The RS400 series of stereo-based RealSense devices features a depth-from-stereo ASIC that supports an Advanced Mode. This mode provides an extended set of controls for fine-tuning the depth generation process and color correction to adapt to different optics, shutter types (global vs. rolling), projector capabilities, and lighting conditions.

    Important Safety and Performance Notes:

    • Safety: The hardware APIs are designed to be safe; you cannot brick the device and can always exit advanced mode to return to default operation.
    • Risk: Tinkering with advanced controls may impact depth quality and stream frame rates. Changes are made at the user's own risk.
  5. Overview of Intel® RealSense™ Camera Tools

    master
    The librealsense repository provides several specialized tools for interacting with, configuring, and debugging Intel® RealSense™ Depth cameras and modules. These tools range from graphical viewers to command-line utilities for data conversion and device enumeration.
  6. Overview of RealSense DDS (RealDDS)

    master

    RealSense DDS (RealDDS) is an abstraction layer built on top of the standard Data Distribution Service (DDS) protocol. It introduces RealSense-specific constructs to manage devices over a network using a publish-subscribe pattern.

    Key characteristics:

    • Self-contained API: It is not reliant on librealsense; instead, librealsense relies on RealDDS. It can be used independently for testing, debugging, or standalone applications.
    • Middleware: It currently uses FastDDS as the underlying implementation.
    • Python Support: A Python module named pyrealdds is available for interacting with the system.
    • Discovery: Clients and servers find each other via a device-level discovery mechanism called device-info, which is broadcast on a topic with a fixed name.
  7. Overview of Intel® RealSense™ Viewer

    master

    RealSense Viewer is a cross-platform UI tool used to access RealSense camera functionality. Key capabilities include:

    • Multi-device streaming: Stream from multiple RealSense devices simultaneously.
    • Pointcloud exploration: Explore pointcloud data in real-time or export it to a file.
    • Recording and Playback: Record RealSense data and play back recorded files.
    • Camera Controls: Access most camera-specific controls, including 3D-generation ASIC registers (when available).
  8. Overview of rs-pointcloud-stitching

    master
    The rs-pointcloud-stitching sample allows you to stitch together two RealSense devices into a single, wide Field of View (FOV) virtual device. It projects the pointclouds from two separate devices onto the same virtual device's image to create a stitched image that adheres to the pinhole camera model. The resulting virtual device can be loaded and viewed using the realsense-viewer application.
  9. Overview of PCL Samples for RealSense

    master

    The PCL Samples demonstrate how to integrate RealSense cameras with the Point-Cloud Library (PCL). These samples complement the standard SDK examples by providing point-cloud processing capabilities.

    Available samples:

    • PCL: A minimal point-cloud viewer that includes PCL processing.
    • PCL-COLOR: A point-cloud viewer that includes RGB PCL processing.
  10. Understand RealDDS Notifications

    master

    RealDDS uses a specific topic to deliver flexible messages from a server to all connected clients via <topic-root>/notification.

    Key Characteristics:

    • Format: Notifications must be a JSON (or possibly CBOR) object. They must include an id field.
    • Broadcast: Messages are broadcast to all clients, not targeted to specific ones.
    • Extensibility: Clients ignore unrecognized id fields or unrecognized fields within a known id. This allows servers to add extra content without breaking clients.
    • QoS (Quality of Service):
      • Reliability: RELIABLE
      • Durability: VOLATILE

    Required Structure:

    {
        "id": "some-message-id",
        "message": "this is a field value"
    }
  11. Use RealSense cameras with OpenCV DNN

    master
    The rs-dnn sample demonstrates how to integrate RealSense cameras with Deep Neural Network (DNN) algorithms using OpenCV. It uses a Caffe model (specifically MobileNet-SSD) to classify objects within the RGB image stream. Once an object is detected in the RGB frame, the demo utilizes the RealSense depth data to calculate the approximate distance to that object.