Traditional Bengali Food Classification and Calorie Measurement Using Machine Learning

Food classification and calorie measurement are turning into fundamental concepts that researchers are applying various techniques in different cuisines to analyze dietary patterns, investigate the nutritional composition of foods, and to calculate total caloric intake. This research paper presents a comprehensive approach for classifying and measuring the calorie content of traditional Bengali dishes using machine learning techniques. A custom dataset containing 2000 images of food items is compiled and annotated for both training and evaluation purposes. YOLOv8 and YOLOv7 architecture are used for both object detection and image segmentation which enable accurate identification and localization of food items within images. Additionally, a mathematical model is created to estimate the volume and calories of the detected food items. The proposed method utilizes the power of deep learning algorithms to effectively classify different Bengali dishes, facilitating their recognition and differentiation. Using YOLOv8, the system is able to achieve robust object detection and segmentation, accurately separating various food items within images. As people are becoming more health-conscious, accurate assessment of dietary caloric intake is becoming significant and promising for leading a healthy life. Our research provides a solution for measuring the caloric intake of the food items individuals consume. Our proposed datasets and models which are trained and evaluated for calculating volume and calorie measurements have the ability of achieving a high level of accuracy and performance. This research provides a valuable foundation for the development of a practical and efficient system for calorie measurement of traditional Bengali dishes, potentially contributing to healthy eating habits and facilitating diet planning.