README.md

Group photo selector and best face detector

This program select the best photo in a set of similar group photos on base with emotions analysis. Using the open source library DeepFace to detect the emotions of the people and select the photo with the most positive emotions. Also, search for the best face in all photos and provide a JSON file with the coordinates of the face and name files to merge in post-processing with Photoshop.

Features

• Select the best photo in a set of similar group photos on base with emotions analysis.
• Search for the best face in all photos and provide a JSON file with the coordinates of the face and name files to crop from the source and paste in the best photo using Photoshop script.

Example

Photo 1	Photo 2	Photo 3	Photo 4	Photo 5	Best photo

This program is oriented for scholar group photos but I am not authorized to share the photos. The photos in the example comes from YouTube.

Target photo

Requirements

• Python 3.7 or later
• OpenCV
• DeepFace
• Numpy
• Pandas
• Matplotlib

Usage

Best photo selector

• Jupyter notebook: panoramic_face_analyze

Photoshop script

1. Edit the file panoramic_face_analyze and change the variable layer2Copy with the name of the layer to copy the faces.
2. Open all the photos in Photoshop.
3. Go to File > Scripts > Browse...
4. Select the file face_crop4paste.jsx and click Open
5. Select the JSON file with the coordinates.