Rename Photos with OCR

This script automates the renaming of specimen photos by extracting labels using Optical Character Recognition (OCR). It uses PaddleOCR to detect text and OpenCV to correct image rotation before processing.

Features

• Automatically extracts text labels from images using OCR
• Corrects rotated images before text extraction
• Replaces spaces and invalid filename characters in extracted text
• Handles duplicate filenames by appending sequential numbers
• Copies and renames images into a separate folder

Caveats

• OCR may not recognise "_" well. Sometimes it is omitted, sometimes it is recognised as other character such as ".", "/". Manual inspection of extracted label is advised.
• The script was used to extract printed text labels from image and not handwriting.

Installation

Prerequisites

Make sure you have Python 3.8 or higher installed. Then, install the required dependencies:

pip install -r requirements.txt

Or manually install the required libraries:

pip install paddleocr opencv-python numpy

The script was written in Python 3.11.

Usage

1. Prepare Your Images

• Place the images you want to rename inside the photos/ folder.

2. Run the Script

Run the script using:

python src/rename_images.py

3. Check the Renamed Files

• The renamed images will be saved inside the renamed/ folder.
• The filenames will be based on the extracted labels.
• If duplicate labels exist, sequential numbers (e.g., _1, _2) will be appended.

Folder Structure

rename-photos-ocr/
│── photos/               # Input images (place your images here)
│── renamed/              # Output directory (renamed images)
│── src/                  # Source code directory
│   ├── rename_images.py  # Main script
│── requirements.txt      # Dependencies list
│── README.md             # Project documentation
│── .gitignore            # Ignore unnecessary files

Example Output

Before:

File was photos/IMG_2137.JPG

After OCR and renaming:

renamed/AAV3FF_00183.jpg

License

This project is licensed under the MIT License.

Acknowledgments

Developed with assistance from ChatGPT.

This work is part of the ADVANCE project funded by BELSPO (BELSPO project RT/23/ADVANCE).