INFORM-Project

Implementation of Food Image Detection and Segmentation methods for edge devices.

Setting Up Your Environment

To run the scripts, you need a properly configured Python environment. Follow these steps to set up:

1. Create and Activate a Virtual Environment:

   For Conda:

   conda create -n myenv python=3.8
   conda activate myenv

   For venv:

   # Create a new virtual environment
   python -m venv myenv
   
   # Activate the environment
   # On macOS/Linux:
   source myenv/bin/activate
   # On Windows:
   myenv\Scripts\activate

2. Install Dependencies: Ensure you have Python installed, along with the following packages:

• OpenCV (opencv-python)
• NumPy
• Scikit-Learn
• Scikit-Image
• PIL (Pillow)
• Matplotlib
• PyTorch
• YOLO (Ultralytics)
• EdgeSAM
• MobileSAM

You can install all required packages using the following command:

pip install -r requirements.txt

Installation

Ensure your environment is active, and you are in the project directory. Clone the repository to your local machine:

git clone https://github.com/inesruizblach/INFORM-Project.git
cd INFORM-Project

Usage

This project includes two main functions for image processing:

1. YOLO EdgeSAM Predictor

   • This function integrates the YOLOv8 detection model with EdgeSAM or MobileSAM segmentation models.

2. YOLO Detect and Segment

   • This function uses the YOLOv8 model for both detection and segmentation.

YOLO EdgeSAM Predictor

To run the YOLO EdgeSAM Predictor, use the following command:

python yolov8-sam.py <img_dir> <ann_dir> <categories_txt_file_path> <sam_model> <yolo_model_path> [--batch_size <int>] [--plot_results]

Example:

python yolov8-sam.py images/ annotations/ categories.txt edgesam models/yolov8_weights.pth --batch_size 200 --plot_results

YOLO Detect and Segment

To run the YOLO Detect and Segment function, use the following command:

python yolov8-segment.py <img_dir> <ann_dir> <categories_txt_file_path> <yolov8_model_path> [--batch_size <int>] [--plot]

Example:

python yolov8-segment.py images/ annotations/ categories.txt models/yolov8_weights.pth --batch_size 200 --plot

Replace the paths with your actual data and model paths. Use the --plot option to enable plotting if the script supports it.

Deactivating the Environment

After you finish running your scripts, deactivate the virtual environment:

deactivate

Poster