Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Feb 20;14(1):4195.
doi: 10.1038/s41598-024-54877-1.

Accelerating computer vision-based human identification through the integration of deep learning-based age estimation from 2 to 89 years

Andreas Heinrich  1
Affiliations

Accelerating computer vision-based human identification through the integration of deep learning-based age estimation from 2 to 89 years

Andreas Heinrich. Sci Rep. .

Abstract

Computer Vision (CV)-based human identification using orthopantomograms (OPGs) has the potential to identify unknown deceased individuals by comparing postmortem OPGs with a comprehensive antemortem CV database. However, the growing size of the CV database leads to longer processing times. This study aims to develop a standardized and reliable Convolutional Neural Network (CNN) for age estimation using OPGs and integrate it into the CV-based human identification process. The CNN was trained on 50,000 OPGs, each labeled with ages ranging from 2 to 89 years. Testing included three postmortem OPGs, 10,779 antemortem OPGs, and an additional set of 70 OPGs within the context of CV-based human identification. Integrating the CNN for age estimation into CV-based human identification process resulted in a substantial reduction of up to 96% in processing time for a CV database containing 105,251 entries. Age estimation accuracy varied between postmortem and antemortem OPGs, with a mean absolute error (MAE) of 2.76 ± 2.67 years and 3.26 ± 3.06 years across all ages, as well as 3.69 ± 3.14 years for an additional 70 OPGs. In conclusion, the incorporation of a CNN for age estimation in the CV-based human identification process significantly reduces processing time while delivering reliable results.

PubMed Disclaimer

Conflict of interest statement

The author declares no competing interests.

Figures

Figure 1
Figure 1
A simplified overview of the C +  + software for CV-based human identification is as follows: Both ante- and postmortem OPGs undergo image processing, involving the following steps. (A) The starting point is the original OPG. (B) This step includes color depth adjustment, border cropping, edge enhancement using eight 3 × 3 Sobel filters, and image noise reduction through an averaging filter. (C) Subsequently, CV features are extracted. The CV features of antemortem OPGs are then stored in the CV database, while the CV features of postmortem OPGs are compared against the CV database, optionally with or without CNN-based age estimation. This process results in matching points between a database entry and the postmortem OPG. If the result with the most matching points ("best result") corresponds to the searched individual, then the identification is considered successful. Additionally, it's possible to analyze, for example, the top 10 results. Dashed lines indicate optional supplementary information.
Figure 2
Figure 2
A scatterplot shows the actual and predicted ages for CNN-based age estimation. The dashed black line represents the ideal case, and the red line represents linear regression. The red dots correspond to the OPGs in Fig. 4M–O.
Figure 3
Figure 3
Boxplots display the signed error between actual and predicted age, along with 95% confidence intervals (indicated by red points), for all age groups in CNN-based age estimation. These boxplots illustrate data variability, accounting for factors that can affect age estimation precision, such as measurement errors, image quality, and clinical considerations like pre- and post-surgical imaging. Narrow confidence intervals indicate consistent and reliable age estimations.
Figure 4
Figure 4
Examples of successful predicted and actual ages for postmortem (A–C) and antemortem OPGs (D–L). Additionally, examples of predicted and actual ages for antemortem OPGs with an absolute error exceeding 20 years (M–O) are marked with red dots in Fig. 2. The OPGs are overlaid with a Grad-CAM representation to highlight the areas of interest, revealing the CNN's focus on specific features important for age estimation.
Figure 5
Figure 5
Overview of the used OPGs, divided by age and sex, for training and validation of the CNN (50,000 OPGs), as well as for an application of the CNN (three postmortem and 10,779 antemortem OPGs). The data is represented through stacked bar charts.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. INTERPOL. Disaster victim identification guide 2018. (Diakses).
    1. Rötzscher K. Forensische Zahnmedizin: Forensische Odonto-Stomatologie. Springer-Verlag; 2013.
    1. Heinrich A, Güttler FV, Schenkl S, Wagner R, Teichgräber UKM. Automatic human identification based on dental X-ray radiographs using computer vision. Sci. Rep. 2020;10:3801. doi: 10.1038/s41598-020-60817-6. - DOI - PMC - PubMed
    1. Heinrich A, et al. Forensic odontology: Automatic identification of persons comparing antemortem and postmortem panoramic radiographs using computer vision. RoFo Fortschritte auf dem Gebiete der Rontgenstrahlen und der Nuklearmedizin. 2018;190:1152–1158. doi: 10.1055/a-0632-4744. - DOI - PubMed
    1. Aliyev, R., Arslanoglu, E., Yasa, Y. & Oktay, A. B. Age Estimation from Pediatric Panoramic Dental Images with CNNs and LightGBM. 2022 Medical Technologies Congress (TIPTEKNO), 1–4 (2022).