Regional Food Recognition with Nutritional Analysis: Attention Mechanism Implementation for Indonesian Cuisine

DOI

10.2991/978-94-6463-924-7_16How to use a DOI?

Keywords

Attention Mechanism; Food Recognition; Indonesian Cuisine; Mobile Deployment; Multi-Task Learning; Nutritional Analysis

Abstract

Indonesia faces two major challenges: the preservation of its food heritage—at risk of digital extinction with 5,350 traditional recipes—and growing public health issues such as diabetes, which increased from 6.9% to 10.9% between 2013 and 2018. Tools that support culturally relevant food tracking are therefore urgently needed. This study has two main aims: (1) to evaluate whether a Convolutional Block Attention Module (CBAM) can improve food classification accuracy for Indonesian dishes, and (2) to assess whether a multi-task architecture can accurately predict nutritional categories using visual features alone. We trained and compared four EfficientNet-B0–based models (Vanilla, CBAM Single, Multi-Task, CBAM Multi-Task) using 815 images across 10 Indonesian food classes. Performance was evaluated through classification accuracy and nutritional prediction F1-score. Contrary to expectations, the CBAM Single-Task model (83.0% accuracy) did not outperform the baseline (85.0%). The best food classification result was achieved by the Multi-Task baseline (88.0%). For nutrition prediction, the CBAM Multi-Task model achieved the highest performance (91.1% F1-score). All models remained computationally efficient, with 3.6–4.0M parameters. The findings reveal a trade-off: multi-task learning provided strong regularization for improving food recognition, while CBAM was more effective when applied specifically to nutritional analysis. The high F1-score demonstrates the feasibility of visual-only nutrition prediction without relying on external databases. This study establishes competitive benchmarks for Indonesian food recognition (88.0%) and visual-based nutritional analysis (91.1%). The integrated framework shows strong potential for deployment in mobile health applications that support public health monitoring and cultural preservation in Indonesia.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Ahmad Fauzan
AU  - Hedy Pamungkas
PY  - 2025
DA  - 2025/12/16
TI  - Regional Food Recognition with Nutritional Analysis: Attention Mechanism Implementation for Indonesian Cuisine
BT  - Proceedings of the 8th International Conference on Informatics, Engineering, Science & Technology (INCITEST 2025)
PB  - Atlantis Press
SP  - 186
EP  - 200
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6463-924-7_16
DO  - 10.2991/978-94-6463-924-7_16
ID  - Fauzan2025
ER  -