Authors:
Anna-Maria Nau、Phillip Ditto、Dawnie Wolfe Steadman、Audris Mockus
Paper:
https://arxiv.org/abs/2408.10414
Automating Human Stage of Decay Identification Using AI: A Detailed Analysis
Introduction
Determining the stage of decomposition (SOD) is a critical task in forensic science, essential for estimating the postmortem interval (PMI) and identifying human remains. Traditionally, this task has been performed manually by forensic experts using visual assessments, which are subjective and labor-intensive. This study, titled “Towards Automation of Human Stage of Decay Identification: An Artificial Intelligence Approach,” explores the feasibility of automating this process using artificial intelligence (AI).
Related Work
Manual Scoring Methods
Two prominent methods for scoring human decomposition are those proposed by Megyesi et al. and Gelderman et al. These methods categorize decomposition into stages based on visual assessments of different anatomical regions, such as the head, torso, and limbs. However, these methods are subjective and prone to human error.
AI in Forensic Science
AI, particularly deep learning models, has shown promise in automating tasks that require human intelligence. Convolutional Neural Networks (CNNs) have been successfully applied in various fields, including medical imaging and forensic science, to analyze large datasets and uncover patterns that might elude human perception.
Research Methodology
Objectives
The primary objective of this study is to evaluate the feasibility of automating the SOD identification task using AI. Specifically, the study aims to:
- Train CNN models to classify SOD for different anatomical regions.
- Assess the reliability of these models compared to human forensic examiners.
Scoring Methods
The study focuses on automating two established human decomposition scoring methods:
- Megyesi et al.: Categorizes decomposition into four stages: fresh, early decomposition, advanced decomposition, and skeletonization.
- **Gelderman et
