The You Only Look Once (YOLO) framework represents a seminal advancement in real-time object detection, widely implemented across domains such as autonomous navigation, surveillance, and robotics.

The latest iteration, YOLOv12, significantly improves computational efficiency and detection accuracy. This article presents a rigorous, stepwise approach to configuring and deploying YOLOv12 on Linux and Ubuntu systems.

System Prerequisites

Prior to installation, ensure that your system satisfies the following requirements:

1. NVIDIA GPU (Optional): To leverage hardware acceleration, install the appropriate NVIDIA driver and CUDA (11.8 or later).
2. Git: Install Git if it is not already present:

sudo apt install git

3. Python 3.11 or Later: Verify installation using:

python3 --version

Installation Procedure

1. System Preparation

Execute the following commands to install necessary system dependencies:

Install NVIDIA GPU Driver and CUDA (Optional): Refer to the NVIDIA Developer Portal for platform-specific instructions.

Install Git:

sudo apt install git

Install Python 3.11 or Later:

sudo apt update
sudo apt install python3.11

2. Repository Acquisition

Clone the official YOLOv12 repository from GitHub:

 git clone https://github.com/sunsmarterjie/yolov12.git
 cd yolov12

3. Establishing a Virtual Environment

Isolating dependencies within a virtual environment prevents conflicts and enhances reproducibility.

Using venv:

python3.11 -m venv yolov12-env
source yolov12-env/bin/activate

Using Conda:

conda create -n yolov12 python=3.11
conda activate yolov12

4. Dependency Installation

Install all requisite dependencies, including PyTorch and FlashAttention:

wget https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.3/flash_attn-2.7.3+cu11torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl
pip install flash_attn-2.7.3+cu11torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl
pip install -r requirements.txt
pip install -e .

5. Validation of Installation

Confirm the integrity of the installation with:

python -c "from ultralytics import YOLO; print('YOLOv12 installed successfully')"

Executing YOLOv12

Once configured, YOLOv12 can be deployed for object detection tasks.

Sample Implementation

The following Python script demonstrates inference on an image:

from ultralytics import YOLO

# Load the trained model
model = YOLO("path/to/model.pt")  # Utilize a pre-trained model

# Perform inference
results = model("path/to/image.jpg")

# Post-process results
for result in results:
    boxes = result.boxes  # Bounding boxes
    probs = result.probs  # Confidence scores
    # result.show()  # Visualize output
    result.save("output.jpg")  # Store processed image

Troubleshooting Strategies

• Dependency Conflicts: Ensure the virtual environment is active and dependencies are properly installed.
• CUDA/Driver Issues: If utilizing a GPU, validate that the NVIDIA driver and CUDA toolkit are correctly installed.

Optimization Recommendations

• Hardware Acceleration: Utilize a GPU to enhance inference speed.
• Custom Dataset Training: Refer to the official documentation for guidelines on fine-tuning YOLOv12 on bespoke datasets.

Future Enhancements

Given the iterative progression of YOLO models, subsequent versions are anticipated to offer superior efficiency and feature enhancements. Maintaining an updated computational environment, including Python and CUDA, ensures continued compatibility and optimal performance.

Appendices

Appendix A: Essential Commands

Appendix B: Troubleshooting Checklist

1. Python Compatibility: Confirm installation of Python 3.11 or newer.
2. Git Availability: Verify Git installation.
3. CUDA Configuration: Ensure correct setup of NVIDIA drivers and CUDA toolkit.
4. Dependency Issues: Activate the virtual environment and reinstall dependencies if discrepancies arise.

Final Considerations

YOLOv12 embodies state-of-the-art advancements in real-time object detection. By adhering to the procedural framework outlined in this guide, researchers and developers can efficiently establish and utilize YOLOv12 within Linux and Ubuntu environments, facilitating its deployment in diverse real-world applications.

Conclusion

Deploying YOLOv12 on Linux and Ubuntu involves methodical steps, including configuring the environment, acquiring the repository, setting up a virtual environment, installing dependencies, and performing verification.

References

1. Run DeepSeek Janus-Pro 7B on Mac: A Comprehensive Guide Using ComfyUI
2. Run DeepSeek Janus-Pro 7B on Mac: Step-by-Step Guide
3. Run Microsoft OmniParser V2 on Ubuntu : Step by Step Installation Guide