We are pleased to announce that our researchers — Aknur Karabay, Prof. Huseyin Atakan Varol, and Prof. Mei Yen Chan — have published a significant new research paper in Scientific Reports (Q1, Top 10%), titled “Improved Food Image Recognition by Leveraging Deep Learning and Data-Driven Methods with an Application to the Central Asian Food Scene.”
This work represents the first to introduce the Central Asian Food Scenes Dataset, a large-scale, high-quality resource featuring 239 food classes and over 69,000 annotated food instances across 21,306 images. Designed to reflect realistic dining contexts common in Central Asia, East Asia, Southeast Asia, and the Middle East, the dataset focuses on complex meal scenes containing multiple food items—moving beyond the limitations of existing datasets which often handle only single-item classification.
Motivated by the rising burden of diet-related diseases in Central Asia, the study addresses a major gap in food computing by advancing computer vision techniques for real-world food localization and detection. This contribution provides a valuable foundation for developing AI-powered tools in digital food journaling, smart restaurants, and health tracking applications.
To read the full paper please visit Scientific Reports.