Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Automation of Human Stage of Decay Identification: An Artificial Intelligence Approach

Anna-Maria Nau, Phillip Ditto, Dawnie Wolfe Steadman, Audris Mockus

TL;DR

The paper addresses automating human stage-of-decay identification to support PMI estimation and forensic workflows. It evaluates two CNN architectures, Inception V3 and Xception, trained via two-step transfer learning on a large, region-labeled decomposition image dataset and conducts an interrater reliability study comparing AI and human raters. Findings show Xception achieved the strongest performance, with head and torso near 0.88 and limbs around 0.70–0.76, and AI–human agreement comparable to human–human for Megyesi and moderate for Gelderman, supporting AI feasibility. The results demonstrate proof-of-concept for AI-assisted SOD classification on archival image collections, while noting limitations such as labeling bias, climate specificity, and the need for larger, balanced datasets and domain-aware labeling.

Abstract

Determining the stage of decomposition (SOD) is crucial for estimating the postmortem interval and identifying human remains. Currently, labor-intensive manual scoring methods are used for this purpose, but they are subjective and do not scale for the emerging large-scale archival collections of human decomposition photos. This study explores the feasibility of automating two common human decomposition scoring methods proposed by Megyesi and Gelderman using artificial intelligence (AI). We evaluated two popular deep learning models, Inception V3 and Xception, by training them on a large dataset of human decomposition images to classify the SOD for different anatomical regions, including the head, torso, and limbs. Additionally, an interrater study was conducted to assess the reliability of the AI models compared to human forensic examiners for SOD identification. The Xception model achieved the best classification performance, with macro-averaged F1 scores of .878, .881, and .702 for the head, torso, and limbs when predicting Megyesi's SODs, and .872, .875, and .76 for the head, torso, and limbs when predicting Gelderman's SODs. The interrater study results supported AI's ability to determine the SOD at a reliability level comparable to a human expert. This work demonstrates the potential of AI models trained on a large dataset of human decomposition images to automate SOD identification.

Towards Automation of Human Stage of Decay Identification: An Artificial Intelligence Approach

TL;DR

The paper addresses automating human stage-of-decay identification to support PMI estimation and forensic workflows. It evaluates two CNN architectures, Inception V3 and Xception, trained via two-step transfer learning on a large, region-labeled decomposition image dataset and conducts an interrater reliability study comparing AI and human raters. Findings show Xception achieved the strongest performance, with head and torso near 0.88 and limbs around 0.70–0.76, and AI–human agreement comparable to human–human for Megyesi and moderate for Gelderman, supporting AI feasibility. The results demonstrate proof-of-concept for AI-assisted SOD classification on archival image collections, while noting limitations such as labeling bias, climate specificity, and the need for larger, balanced datasets and domain-aware labeling.

Abstract

Determining the stage of decomposition (SOD) is crucial for estimating the postmortem interval and identifying human remains. Currently, labor-intensive manual scoring methods are used for this purpose, but they are subjective and do not scale for the emerging large-scale archival collections of human decomposition photos. This study explores the feasibility of automating two common human decomposition scoring methods proposed by Megyesi and Gelderman using artificial intelligence (AI). We evaluated two popular deep learning models, Inception V3 and Xception, by training them on a large dataset of human decomposition images to classify the SOD for different anatomical regions, including the head, torso, and limbs. Additionally, an interrater study was conducted to assess the reliability of the AI models compared to human forensic examiners for SOD identification. The Xception model achieved the best classification performance, with macro-averaged F1 scores of .878, .881, and .702 for the head, torso, and limbs when predicting Megyesi's SODs, and .872, .875, and .76 for the head, torso, and limbs when predicting Gelderman's SODs. The interrater study results supported AI's ability to determine the SOD at a reliability level comparable to a human expert. This work demonstrates the potential of AI models trained on a large dataset of human decomposition images to automate SOD identification.
Paper Structure (12 sections, 5 equations, 2 figures, 6 tables)

This paper contains 12 sections, 5 equations, 2 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Materials and methods
  3. Stage of decay scoring methods
  4. The human decomposition dataset
  5. Data processing and labeling
  6. AI model development
  7. Interrater test
  8. Results
  9. SOD classification
  10. Interrater test
  11. Discussion
  12. Conclusion

Figures (2)

  • Figure 1: The data processing pipeline.
  • Figure 2: The SOD classification framework was trained using a two-step transfer learning approach. In step 1, the pre-trained convolutional layers of the base model (blue rectangle) were frozen, and only the newly added classifier layers (yellow rectangle) were trained. In step 2, the previously frozen layers from step 1 (blue rectangle) were unfrozen, and the entire model was trained end-to-end.