Welcome to ForensicFit Documentation!#
ForensicFit is a Python package designed to preprocess scanned images from various sources and generate a database to be used in different machine learning approaches. This package prepares the data using four distinct techniques, which will be explained in the tutorial sections. ForensicFit leverages state-of-the-art image processing methods to analyze and store the generated data, ensuring compatibility with popular machine learning packages such as TensorFlow, PyTorch, and SciKit-learn. It utilizes NumPy, SciPy, matplotlib, OpenCV, scikit-image, PyMongo, and GridFS. For ease of use and future development, the package adheres to PEP-257 and PEP-484 for documentation and type hints, respectively.
Package Structure#
ForensicFit is organized into three main sub-packages: core, database, and utils.
core: This sub-package contains the essential functionalities of ForensicFit, including Python classes that manage read/write, analysis, and metadata storage. These classes provide a data structure skeleton for the package and define standards for future implementations related to different types of materials.database: This sub-package offers an efficient and flexible method for storing and retrieving raw and preprocessed data. Although the rest of the package does not depend on this sub-package, it has been included to simplify the data storage and query process. Users can still store and access raw or analyzed data using traditional image storage methods.utils: Theutilssub-package contains various image manipulation, plotting, and memory access tools used throughout the package.
Installation#
To install ForensicFit, use the following command:
pip install forensicfit
Quick Start#
Here’s a quick example of how to use ForensicFit:
>>> import forensicfit as ff
# Load your image file
>>> path = 'path/to/LQ-HT-1.jpg'
>>> tape = ff.core.Tape.from_file(path)
>>> print(tape)
Mode: material
Resolution: (2471, 6289, 3)
Path: path/to/LQ-HT-1.jpg/LQ_099.tif
Filename: LQ_099.tif
Compression: raw
DPI: (1200.0, 1200.0)
Flip horizontal: False
Flip vertical: False
Split vertical: {'side': None, 'pixel_index': None}
Label: None
Material: tape
Surface: None
Stretched: False
Contents: