Publication

Abstract : The modern way of life has greatly escalated the occurrence of chronic illnesses especially the obesity which is one of the leading causes of type 2 diabetes, cardiovascular diseases and hypertension. Calories or energy content is one of the main factors that increase the risk of obesity, and it is usually given in terms of how much food you intake daily. Although pharmaceutical interventions may reduce an appetite, it is associated with unpleasant side effects. Hence, a more sustainable solution means having a balanced diet and engaging in physical exercise and paying close attention to the amount of calories that one consumes as suggested by nutritionists. Against this backdrop, there has been a significant focus on development of systems that can estimate the content of calories of food based on the image of the food. The idea behind such systems is to scan foods that have high amounts of energy, especially those that are high in fats and sugars, to help individuals control their food intakes. It starts by taking pictures of food products, which are then categorized based on highly developed deep learning algorithms, including convolutional neural networks (CNNs) such as MobileNetV2. After proper identification of the food item, the value of the calories can be approximated. The accuracy of calorie estimation greatly depends on the accuracy of the image classification model. This involves training the model on large datasets and critically testing it with a variety of and difficult test cases. The suggested solution emphasizes the opportunities of artificial intelligence to aid healthier lifestyle decisions by automated and trusted dietary tracking.