Dietary Monitoring with Deep Learning and Computer Vision
Swati Khawate, Swapnali Gaikwad, Yatri Davda, Radha Shirbhate, Poonam Suraj Gham, Vishal Borate
Abstract
In an era where maintaining a nutritious diet is paramount for optimizing well-being amidst the fast pace of life, this research leverages cutting-edge deep learning and computer vision methodologies to revolutionize food photo analysis. By integrating neural networks and image recognition, our method demonstrates exceptional proficiency in accurately identifying and categorizing common food items, enabling precise calculation of their caloric and nutrient composition directly from images. The foundation of our approach lies in a meticulously curated dataset, showcasing the versatility of our methodology across a wide range of food types. Following rigorous evaluation, Model-44 emerges as the top performer, with an impressive 89.6% training accuracy and 82.8% validation accuracy. Real-world trials confirm its practical efficacy in instantly recognizing food images, promising diverse applications across various domains. The ability to swiftly and accurately assess nutritional attributes through image analysis presents novel opportunities for dietary monitoring, empowering individuals to make well-informed nutritional choices effortlessly. Amidst growing health consciousness, our research signifies a significant leap forward in simplifying calorie tracking and promoting informed dietary habits, heralding a healthier future.