UltraDarkFusion 4.0 – Next-Level Object Detection & Segmentation

UltraDarkFusion 4.0 is an advanced GUI for object detection, segmentation, and computer vision, designed to streamline dataset management, annotation, model training, and real-time inference.

What's New in UltraDarkFusion 4.0?

• Full Segmentation Support – segmentation labeling, training, and inference.
• Improved Dataset Management – Enhanced tools for dataset preparation and duplicate removal.
• Optimized Model Training – Enhanced performance with YOLO and SAM integration.
• TensorRT Inference – Accelerate inference with TensorRT engine models.
• Expanded Analytics – Detailed metrics for tracking model performance.

Key Features

• Advanced Labeling – Bounding box and segmentation annotations compatible with YOLO.
• Video Processing – Extract frames and process YouTube videos for dataset generation.
• Customizable UI & Themes – Personalize your interface and workflow.
• Automatic Labeling – Quickly annotate datasets using pre-trained weights.

---

🚀 Installation Guide

Required Build Tools

Open Command Prompt and run:

winget install git.git
winget install Kitware.CMake

Visual Studio 2019

• Download and install Visual Studio 2019 Community Edition
• Ensure Desktop Development with C++ and Python Development workloads are selected.

Install CUDA and cuDNN

• Install CUDA 12.6
• Download cuDNN compatible with CUDA 12.6 from NVIDIA Developer
• Extract and copy cuDNN files to:

  C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.6
  

Clone UltraDarkFusion Repository and OpenCV

cd C:\
git clone https://github.com/lordofkillz/DarkFusion.git
cd DarkFusion
git clone --branch 4.9.0 https://github.com/opencv/opencv.git opencv-4.9.0
git clone --branch 4.9.0 https://github.com/opencv/opencv_contrib.git opencv_contrib-4.9.0
mkdir build install

Mandatory Downloads

• SAM Weights and Files
  • Extract to C:\DarkFusion\UltraDarkFusion

Anaconda Setup

• Download and install Anaconda.
• Accept default installation settings.

Setup Environment

• Run fusion_install.bat as administrator to create and configure the environment:

cd C:\DarkFusion\
fusion_install.bat

Compile OpenCV with CUDA (Optional)

To compile OpenCV with CUDA acceleration (recommended for improved performance):

cd C:\DarkFusion\
fusion_cuda.bat

Install Darknet

• Install Darknet by Hank-AI.

---

Community and Support

• UltraDarkFusion Discord: Join Discord

Acknowledgments

Special thanks to these open-source projects:

• SAM (Segment Anything)
• GroundingDINO
• Ultralytics YOLO
• SAHI

Inspired by DarkMark, developed by Stephane from Hank-AI.

Goals for UltraDarkFusion

• Remove OpenCV DNN and create a Python wrapper for Darknet #in development
• Add support for OBB (Oriented Bounding Boxes) and pose estimation
• Remove the document-based help and train a chatbot for assistance
• Improve code efficiency (Ongoing...)
• Upload UltraDarkFusion to PyPI for easier installation and distribution

---