Explaining CNN-Based Body Measurement Estimation in Children Using Grad-RAM
Authors:
- Hezha Hassan MohammedKhanHezha Hassan MohammedKhan
- Eric PostmaEPEric Postma
- Andrew AbelAAAndrew Abel
- Çiçek GüvenÇGÇiçek Güven
- John BurksJBJohn Burks
- Liam BrownLBLiam Brown
- Hein FleurenHFHein Fleuren
Preprint
- 16 February 2026
- preprint
- Published by Elsevier in SSRN Electronic Journal
Abstract
Convolutional Neural Networks (CNNs) can estimate anthropometric measurements from 2D images, yet the opacity of CNN models hinders deployment in health-adjacent settings. We study explainability for regression by adapting and comparing Gradient-weighted Regression Activation Mapping (Grad-RAM) variants to four child measurements (height, weight, waist circumference, head circumference) using a DenseNet-121 backbone and a landmark-guided cropping pipeline. To the best of our knowledge, this is the first paper to apply explainable AI (XAI) methods to body measurement estimation tasks, specifically of children. We evaluate five formulations: (i) Grad-RAM (signed), (ii) a Grad-CAM reference, (iii) GradRAM (|P|), (iv) Grad-RAMP (element-wise, | · | after summation), and(v) Element-wise Grad-RAMP (|·| before summation). Quantitatively, testset MAE reaches 2.80cm (height), 1.30kg (weight), 2.30cm (waist), and 1.08cm (head), matching or improving prior ARAN baselines. Qualitatively, all Grad-RAM/RAMP variants exhibit task-aware attributions (e.g., bodywide focus for height; abdominal emphasis for waist) while also revealing reliance on contextual cues. In a targeted ablation, masking background consistently degraded accuracy across tasks, confirming sensitivity to nonanatomical signals. We use these explanations not merely as post hoc audit tools but to position it in a cycle that feeds back to data collection and model training: saliency maps diagnose dataset and preprocessing biases (e.g., ruler dependence), guide data collection (background diversity), inform augmentation (cropping and aspect-ratio handling), and motivate model choices. This cycle operationalizes Grad-RAM for improving CNN regressors in imagebased body measurement estimation task.Keywords
This publication has 12 references indexed in Scilit:
- ARAN: Age-Restricted Anonymized Dataset of Children Images and Body MeasurementsJournal of Imaging, 2025
- A systematic review of explainable artificial intelligence for spectroscopic agricultural quality assessmentComputers and Electronics in Agriculture, 2025
- Simultaneous measurement of carbon emission and gas temperature via laser-induced breakdown spectroscopy coupled with machine learningOptics Express, 2023
- A comprehensive taxonomy for explainable artificial intelligence: a systematic survey of surveys on methods and conceptsData Mining and Knowledge Discovery, 2023
- Identification of malnutrition and prediction of BMI from facial images using real‐time image processing and machine learningIET Image Processing, 2021
- Ensemble Manifold Regularized Multi-Modal Graph Convolutional Network for Cognitive Ability PredictionIEEE Transactions on Biomedical Engineering, 2021
- Human biases in body measurement estimationEPJ Data Science, 2020
- Explainability for Regression CNN in Fetal Head Circumference Estimation from Ultrasound Images2020
- Visualization Method Corresponding to Regression Problems and Its Application to Deep Learning-Based Gaze Estimation ModelJournal of Advanced Computational Intelligence and Intelligent Informatics, 2020
- Densely Connected Convolutional Networks2017