Achmad Farizul Islam, Muchamad Wahyu Prasetyo, Aripriharta, Titi Mutiara Kiranawati, Mazarina Devi, Satia Nur Maharani, Gwo-Jiun Horng
The increasing prevalence of chronic diseases such as obesity, diabetes, and cardiovascular disorders highlights the importance of accurately and efficiently monitoring diet and nutritional intake. Unfortunately, traditional methods of food identification still rely on manual input that is slow, impractical, and prone to errors, especially for ordinary users. To overcome this challenge, this study developed an integrative system based on a robotic arm with a high-resolution camera and a mobile application to automatically recognize food and estimate its nutritional content in real-time. The system utilizes the Convolutional Neural Network (CNN) algorithm to classify food images captured by a robotic camera, then forwards the classification results to a mobile app to display nutritional information. The dataset contains 2, 1 8 6 images of Indonesian food used for model training and evaluation, equipped with a preprocessing process and data augmentation to improve the model's resilience to visual variation. Test results show that the system can achieve a recognition accuracy of 91% under normal lighting conditions and maintain an accuracy above 85% under extreme lighting conditions. Estimates of nutritional values show a high level of precision when compared to trusted nutrition databases such as USDA and Fat Secret, with low MAPE and RMSE values. The mobile app shows a fast response time (10 seconds on average) and a user satisfaction score of 4. 6 out of 5. This approach shows great potential in daily nutrition monitoring, nutrition education, and therapeutic dietary support. This system is expected to contribute to preventive efforts in public health through smarter, faster, and more accessible food monitoring. © 2025 IEEE.
Universitas Negeri Malang, Department of Electrical Engineering and Informatics, Malang, Indonesia; Universitas Negeri Malang, Department of Culinary and Fashion Education, Malang, Indonesia; Universitas Negeri Malang, Department of Accounting, Malang, Indonesia; Taiwan University of Science and Technology, Department of Electronics Engineering Southern, Taiwan
