Issuu

International Research Journal of Engineering and Technology (IRJET)

e-ISSN: 2395-0056

Volume: 12 Issue: 09 | Sep 2025

p-ISSN: 2395-0072

www.irjet.net

Deep Learning Based Food Nutrition Estimation Using Images Tejashwini D S1, Dr. Rashmi C R2, Dr. Shantala C P3 1PG Student, Dept. Of Computer Science & Engineering, Channabasaveshwara Institute of Technology, Gubbi,

Karnataka, India

2Assistant Professor, Dept. Of Computer Science & Engineering, Channabasaveshwara Institute of Technology,

Gubbi, Karnataka, India

3 Professor & Head, Dept. Of Computer Science & Engineering, Channabasaveshwara Institute of Technology,

Gubbi, Karnataka, India ---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - In the modern day busy lifestyle, people are

Nevertheless, it may be challenging and ineffective to estimate calories manually. Most people use the labels on foods, nutritional charts, or mobile apps, in which they manually enter the food product that they have eaten. This method is not only time consuming but can also be subject to human error particularly where the portions are not of the same size. As an example, the number of calories in a banana can vary not only with a kind of fruit but also with its size and maturity. Also, not all people know how to identify the correct values of calories or measures of portions. This provides a distance between what it would take to be able to track important calorie intake and the convenience of it being done in daily life.

The methodology includes preprocessing the food images by resizing, normalizing, and augmenting images, and classifying them by a fine-tuned MobileNetV2 model. The experimental results demonstrate that the system can be classified accurately at a reasonable level of efficiency with a high degree of classification, and can be applied in real-time. The paper demonstrates that deep learning and computer vision can be used to address the gap between dietary monitoring requirements and practicality, and to deliver an intelligent, scalable, and usable solution to calorie estimation.

In a bid to eliminate these constraints, the proposed project will be to create a smart food-recognition and caloriecounting machine that will rely on artificial intelligence to automate the task. The system receives a food picture as input, recognizes the food item with deep learning methods, and then gives the calorie estimate as a mapping of the identified food to a pre-existing nutritional database. This saves the user a lot of effort involved and enhances precision in tracking diets.

finding it very difficult to properly track what they eat, especially in determining the number of calories they consume. Manual search in nutritional tablets or smart apps is extremely time-consuming, error-prone, and fails to consider the variability of portions, which is a standard procedure. This paper introduces a Food Recognition and Calorie Estimation System based on Deep Learning and automates the task of recognizing food objects on an image and mapping them to nutritional data. It uses the MobileNetV2 convolutional neural network architecture, which is selected due to its lightweight nature and processing capabilities, and it can be deployed on a mobile and embedded platform.

The field of computer vision and deep learning is the key to the existence of this project. In particular, the Convolutional Neural Network (CNN) is applied to identify and label food pictures. CNNs represent a useful sub-type of neural networks that find extensive application in image recognition with their capacity to automatically identify patterns like edges, shapes, textures, and colors that are inherent to data.

Key Words: Food recognition, calorie estimation, deep learning, convolutional neural networks (CNN), MobileNetV2, computer vision, transfer learning, nutritional database, image classification, mobile health applications, dietary monitoring.

1.INTRODUCTION

The CNN in this project is constructed with the MobileNetV2 architecture that is a lightweight and highly accurate deep learning model. MobileNetV2 is especially apt in the real-world due to its computational efficiency, which implies that it can be implemented on mobile devices and embedded systems without the heavy hardware resource usage. This renders the project accurate, practical, and scalable.

The modern world today where lifestyle is becoming increasingly hectic and technology-sensitive is making people more conscious of their health and their dieting. Healthy nutrition is crucial to preserving overall well-being, preventive diseases, and an active lifestyle. Among the elements of nutrition management, the calorie content of food is among the most significant ones. Calories are the measurement of energy, and daily calorie consumption is a practice required by all individuals who are trying to lose weight, control obesity, develop fitness programs, or merely live a healthy life.

Impact Factor value: 8.315

ISO 9001:2008 Certified Journal

Page 27