Scalable AI for Food Recognition: A Novel YOLOv9-Based Framework Implementation

Indian cuisine, with its diverse and intricate compositions, presents unique challenges for automated food recognition systems. This study leverages YOLOv9, enhanced with attention mechanisms and feature fusion techniques, to address these complexities. Using a curated dataset of regional cuisines, the model achieved a mean Average Precision (mAP) of 92.5%, outperforming prior models like YOLOv8. Robust cross-dataset evaluations on benchmarks such as Food-101 highlight its generalizability. This work demonstrates the potential of AI in cultural preservation, dietary monitoring, and culinary applications, offering a scalable framework for food recognition tasks. Indian cuisine, with its unparalleled diversity and complexity, presents a significant challenge for automated food recognition systems. Characterized by thousands of distinct dishes across regions like North, South, East, and West India, each with unique visual, textural, and compositional features, Indian food poses an intricate problem for traditional computer vision models. Recognizing and categorizing these dishes accurately is essential not only for culinary documentation but also for applications such as dietary monitoring, restaurant automation, and cultural preservation.