Moonlight
  • Welcome to Moonlight!
  • Getting Started!
  • GUIDES
    • Arduino Section
      • Arduino Leonardo Section
        • Arduino Leonardo Spoofing Guide
        • Leonardo + Moonlink Guide
      • Arduino Due Section
        • Arduino DUE Spoofing Guide
        • DUE + Moonlink Guide
        • Advanced Arduino Due Troubleshooting
    • OBS/NDI GUIDE
      • Simple Install
      • Advanced Install
      • Direct PC to PC link
    • Capture Card Guide
    • A.I. Install Guide for Darknet
    • Creating Datasets for Moonlight
      • Creating Datasets for YOLOv4 in Darknet
      • Creating Datasets for YOLOv8 +
  • OVERVIEWS
    • Moonlight Color Overview
    • Moonlink Overview
  • ABOUT MOONLIGHT
    • Our Solutions and How They Work
Powered by GitBook
On this page
  • Step 1: Download and Extract the OpenCV Build
  • Step 2: Run the Installer
  • Step 3: Select Your GPU Architecture
  • Step 4: Installation Completion
  • Troubleshooting Common Issues
  1. GUIDES

A.I. Install Guide for Darknet

Follow this guide on your 2nd PC to correctly install OpenCV with CUDA support . Note that this is only needed for older versions of Moonlight which are not recommended.

PreviousCapture Card GuideNextCreating Datasets for Moonlight

Last updated 4 months ago

Step 1: Download and Extract the OpenCV Build

  1. Download the OpenCV build from the provided Google Drive link.

  1. Extract the ZIP file to C:\, ensuring it is not nested inside another folder.

    • The final directory should be: C:\opencv-build

    • If you see C:\opencv-build\opencv-build, move the contents to just C:\opencv-build.

Step 2: Run the Installer

  1. Open File Explorer and navigate to C:\opencv-build.

  2. Locate and run installer(GPU).bat.

    • right-click and select 'Run as administrator'.

    • The installer can take up to 40 minutes to download the dependencies needed to build opencv with cuda support.

Step 3: Select Your GPU Architecture

  1. When prompted, choose the GPU architecture that matches your graphics card.

    • For NVIDIA 40-series GPUs, select the 30-series option instead.

    • The build phase can take up to 40 minutes depending on GPU.

Step 4: Installation Completion

  1. Wait for the installation to finish.

  2. Take a screenshot of the final output and share it as requested. This step is important because it allows us to check if there are any errors in the install.

Troubleshooting Common Issues

Issue: CUDA or cuDNN Failed to Install

If the build phase completes early and the output indicates that CUDA or cuDNN has failed (or shows "NO" next to it), follow these steps:

  1. Open Visual Studio Installer.

  2. Click Resume if it needs to finish installing.

  3. Download CUDA 11.1 from the official NVIDIA site.

  4. After Visual Studio has finished being installed, install CUDA 11.1 following the on-screen instructions.

  5. After installation is complete, go back to C:\opencv-build.

  6. Run build.bat.

  7. Select your GPU architecture when prompted.

  8. The installation should now proceed successfully.

Other Troubleshooting Tips:

  • If you encounter errors, verify that you extracted the folder correctly to C:\opencv-build.

  • If the batch script fails, make sure you were running Command Prompt as Administrator.

https://drive.google.com/file/d/1jDDhIitfYsqWsuaSnRKMi9gdBUxRtspt/view?usp=drive_link