Image Segmentation using YOLOv11 and SAM v2

This repository contains an advanced image segmentation project that leverages the powerful object detection capabilities of YOLOv11 and the state-of-the-art segmentation capabilities of SAM (Segment Anything Model) v2. The project demonstrates how these models can work together to achieve accurate and efficient image segmentation for a variety of use cases.

Features

• Object Detection with YOLOv11: Utilizes the YOLOv11 model for real-time object detection with high precision.
• Segmentation with SAM v2: Applies the Segment Anything Model (SAM) v2 for accurate and flexible segmentation.
• Pipeline Integration: Combines the strengths of YOLOv11 and SAM v2 into a seamless image segmentation pipeline.
• Easy Customization: Modular codebase for adapting to different datasets and use cases.

Applications

• Autonomous driving
• Medical image analysis
• Agricultural monitoring
• Industrial inspection
• Any domain requiring precise object segmentation

Installation

1. Clone the repository:

   git clone https://github.com/sohamvsonar/Image-Segmentation-Using-YOLO-SAM.git
   cd Image-Segmentation-Using-YOLO-SAM

2. Install the required dependencies:

   pip install tensorflow pytorch roboflow

Usage

1. Input Preparation

2. Run the Segmentation Pipeline

• Execute the script to perform object detection and segmentation:

  python main.py

3. View Results

• Segmented images will be saved in the output_images/ directory.
• Visualizations will include bounding boxes (from YOLOv11) and segmentation masks (from SAM v2).

Project Structure

How It Works

1. YOLOv11 Detection: Detects objects in the input image and generates bounding boxes.
2. SAM v2 Segmentation: Extracts segmentation masks for the detected objects using SAM v2.
3. Output Generation: Combines detection and segmentation results for visualization.

Dependencies

• Python 3.8+
• PyTorch 1.12+
• OpenCV
• NumPy
• Matplotlib

See requirements.txt for the full list of dependencies.

Demo

Contributing

Contributions are welcome! Please open an issue or submit a pull request to suggest improvements or report bugs.

Acknowledgements

• YOLOv11: For object detection capabilities.
• Segment Anything Model (SAM) v2: For segmentation capabilities.

Contact

For questions or feedback, please contact sohamvsonar.

---

Feel free to fork this repository and adapt it to your own use cases!