BMI Prediction Using Facial Features with Deep Learning Techniques

Abstract

Body Mass Index (BMI) is a simple measure that links a person’s weight to their height. It’s widely used to assess health levels and risks. But measuring height and weight can be inconvenient. Or not always possible. In this study, we explore an alternative—predicting BMI from facial images.One such approach involves leveraging facial features extracted from images to predict BMI, eliminating the need for direct physical measurements. This study presents a deep learning-based system that utilizes Convolutional Neural Networks (CNNs), particularly ResNet50 architecture, to analyze facial images and predict BMI through a regression layer. The model is trained on pre-processed facial datasets, using Haar cascade classifiers for face detection and standardization. Once the features are extracted, the system classifies the estimated BMI into standard health categories such as underweight, normal, overweight, or obese. This approach offers a non-intrusive, practical alternative for BMI estimation, particularly useful in healthcare applications, mobile health platforms, and wellness tools. Using Convolutional Neural Networks (CNNs) and regression models, we created a system that takes in a face and gives an estimated BMI. No physical measurements. Just one photo. This approach can help in healthcare, social platforms, or any scenario where quick BMI estimation is useful.

Introduction

Introduction:
Body Mass Index (BMI), based on weight and height, is a standard health indicator classifying individuals as underweight, normal, overweight, or obese. However, BMI doesn't differentiate between fat and muscle and requires physical measurements, which may not always be accessible. Sedentary lifestyles are increasingly contributing to higher BMI rates.

Facial Analysis for BMI:
Recent studies reveal a strong correlation between facial structure (e.g., jawline, cheekbones, eyebrow spacing) and BMI. Deep learning enables non-invasive BMI estimation from facial images, offering a practical alternative when weight/height data is unavailable.

Related Research:
Prior studies used various CNN architectures (e.g., VGG, ResNet50, DenseNet, LightCNN) to estimate BMI from facial images. Some models integrated demographic data and sent results to healthcare providers. Public image datasets have supported multi-stage BMI prediction systems.

Proposed Model:
The project introduces a CNN-based model using the ResNet50 architecture trained on ImageNet. Facial images are detected and cropped using Haar cascade classifiers, resized to (224, 224, 3), and passed through the model to predict BMI categories.

System Details:
Developed in Python with libraries like TensorFlow, Keras, OpenCV, and NumPy.
Key components:

• CNN for visual pattern recognition

• ResNet50 for deep feature extraction

• ReLU & Softmax activation functions

• Haar Cascade for facial detection

Implementation:
Images are uploaded via a simple UI, processed for face detection, and passed through the trained model. The output includes an estimated BMI and its classification (Underweight, Healthy, Overweight, Obese).

Results:
Examples:

• Image 1 → BMI: 20.70 → Healthy

• Image 2 → BMI: 17.71 → Underweight

This deep learning system provides a user-friendly, automated way to estimate BMI using just a facial photo.

Conclusion

The findings highlight a measurable relationship between facial structure and BMI. Those with elevated BMI are at greater health risk. The system does not require full-body images—just a face is enough. The process involves feature extraction, model training, and classification. No significant gender bias was found, though performance may still be improved. Main challenges include the need for large datasets and the difficulty in predicting multiple individuals in a single image. Future models can benefit from higher diversity in training data to enhance robustness and accuracy. The model also struggles with translating 2D facial data into precise BMI values.

References

[1] Hera Siddiqui, Ajita Rattani, Dakshina Ranjan Kisku and Tanner Dean; “AI-based BMI Inference from Facial Images: An Application to Weight Monitoring”; 2020. [2] Dhanamjayulu C, Nizhal U N, Praveen Kumar Reddy Maddikunta, Thippa Reddy Gadekallu, Celestine Iwendi, Chuliang Wei, Qin Xin; “Identification of malnutrition and prediction of BMI from facial images using real-time image processing and machine learning; 2021. [3] Enes Kocabey, Mustafa Camurcu, Ferda Ofli , Yusuf Aytar, Javier Marin, Antonio Torralba, Ingmar Weber; “Face-to-BMI: Using Computer Vision to Infer Body Mass Index on Social Media”; 2017. [4] Kaiming He, Xiangyu Zhang, Shaoqing Ren and Jian Sun; “Deep Residual Learning for Image Recognition”; 2015. [5] Lingyun Wen, Guodong Guo; “A computational approach to body mass index prediction from face images”; 2013. [6] Nadeem Yousafa , Sarfaraz Husseinb and Waqas Sultani; “Estimation of BMI from Facial Images using Semantic Segmentation based Region-Aware Pooling”; 2021. [7] Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li and Li Fei-Fei; “ImageNet: A Large-Scale Hierarchical Image Database”; 2009. [8] Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang , Andrej Karpathy , Aditya Khosla, Michael Bernstein , Alexander C. Berg , Li Fei-Fei; “ImageNet Large Scale Visual Recognition Challenge”; 2015.

Copyright

Copyright © 2025 Anasuri Satya Chetan, Dhanush Bonthala, P. Akshay Raj, K. Sree Latha. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.