Image Analysis CLI Tool

A robust, configurable command-line tool written in Python for performing batch analysis on image datasets. This tool recursively scans a directory for image files, extracts key metadata, and saves the results in a structured CSV file. The tool uses parallel processing to speed up analysis on multi-core systems.

Key Features

• Configurable: All key parameters (input folder, output file, file extensions) are managed externally in a JSON file (example: configs/config.json).
• Robust Error Handling: The application is designed to be resilient. It gracefully handles common errors such as missing directories, invalid image files, and file permissions.
• Memory Efficient: Uses a streaming approach (Python generators) to process a large number of images with a very low and constant memory footprint.
• Platform Independent: Built with pathlib to ensure it runs correctly on Windows, macOS, and Linux.
• Clean Architecture: The code is logically separated into distinct, reusable, and testable modules (ConfigReader, FileExtractor, ImageAnalyzer, CSVWriter).
• Parallel Processing: Parallel processing implemented for image analysis to speed up the biggest bottleneck in the pipeline.

Setup and Installation

1. Clone the repository:

   git clone git@github.com:Marenix/image-analysis.git
   cd image-analysis

2. Install dependencies: This project requires the Pillow library and OpenCV for image processing.

   pip install -r requirements.txt

(Note: All other required modules like pathlib and logging are part of the standard Python library.)

Configuration

Before running the application, create your JSON config file in configs/. You can use available config.json file as a template.

{
    "FileExtractor": {
        "input_location": "data/",
        "extensions": [".jpg", ".jpeg", ".png"]
    },

    "CSVWriter": {
        "output_location": "output/image_metadata.csv"
    },

    "AnalyzerSettings": {
        "face_detection_model_path": "models/face_detection_yunet_2023mar.onnx",
        "visualize": false
    }
}

• input_location: The directory containing the images you want to analyze (NOTE: all subdirectories are searched as well).
• extensions: A list of file extensions to include in the scan.
• output_location: The path where the final CSV report will be saved. The directory will be created if it doesn't exist.

How to Run

Execute the main.py script from the root directory of the project, providing the path to your configuration file.

Optional arguments:

• --input_location: Path to the input folder.
• --output_location: Path to the output folder (including the CSV file name).
• --visualize: Bool flag that will save images with bounding boxes around faces where detected if specified

python main.py --config_path configs/config.json

The script will then process the images and generate the CSV file in the location specified in your config.

Example Output

An example of the visual output generated by the tool when running with the '--visualize' flag.

Current Extracted Data

The current version of this tool extracts the following basic metadata:

• filename: The name of the image file.
• filesize: The size of the file in bytes.
• width: The width of the image in pixels.
• height: The height of the image in pixels.
• aspect_ratio: The aspect ratio of the image.
• average_color: The average color of the whole image.
• num_of_faces: The number of faces detected in the image.

Future Work

This project is built to be easily extended. The next planned steps include:

• Perform object detection in the image to detect some common/interesting objects.