This Python script uses the DeepFace library to search for images in a specified directory (or directories) that match one or more reference images. The matching images are then copied to an output folder.

Before running the script, ensure that you have the following dependencies installed:

• Python 3.x
• OpenCV (cv2)
• DeepFace (deepface)

You can install the required Python packages using pip:

pip install opencv-python
pip install deepface

1. Clone or download the repository containing the script.

2. Update the following paths in the script according to your file structure:

   • reference_image_folder: Path to the folder containing the reference images.
   • output_folder: Path to the folder where the identified images will be saved.

3. Place your reference image(s) in the reference_image_folder.

4. In the script, update the directory path(s) to scan for images in the scan_directory() function call. For example:

   scan_directory(r'C:\Users\YourUsername\Pictures\YourFolderName')

   You can provide multiple paths by adding more scan_directory() calls.

5. Run the script using the following command:

   python script_name.py
   

   Replace script_name.py with the actual name of your Python script file.

The script will scan the specified directory (or directories) and compare each image against the reference image(s). If a match is found, the image will be copied to the output_folder while maintaining the original directory structure.

1. The script loads the reference image(s) from the reference_image_folder.
2. It creates the output_folder if it doesn't already exist.
3. The scan_directory() function is called with the path(s) to scan for images.
4. For each image found in the specified directory (or directories), the script performs the following steps:
   • Load the target image using OpenCV (cv2.imread()).
   • Compare the target image against each reference image using the DeepFace.verify() function from the deepface library.
   • If a match is found (i.e., result["verified"] is True), copy the matching image to the output_folder while preserving the original directory structure.
   • Print a message indicating the path of the saved image.
5. If an error occurs during the processing of an image, it prints an error message with the image path and the exception details.

Note: The script uses the VGG-Face model from DeepFace for face recognition. You can modify the model_name parameter in the DeepFace.verify() function to use a different model if desired.

This project is licensed under the MIT License.