Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{208852,
        author = {Bhoomika R and Tejaswini K R and Vindya A P and Thanusha J and Prof. Amulya},
        title = {FoodLens AI: An AI-Powered Food Inventory and Recommendation System Using YOLOv8-Based Food Detection},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {4},
        pages = {3086-3092},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=208852},
        abstract = {Keeping track of what is inside a household refrigerator is still largely a manual task, which makes it hard to know which items are available and how to use them. This paper presents FoodLens AI, a web-based food inventory and recommendation system that uses YOLOv8 object detection to identify food items in refrigerator images and turn the detections into inventory records and recipe suggestions. The system is organized as a React web client that submits images to a FastAPI backend, where a YOLO model performs inference and a PostgreSQL database stores users, inventory, food items, recipes, and recommendation information. A first-generation detector (V1, YOLOv8n) reached an approximate validation precision of 0.825, recall of 0.752, [email protected] of 0.813, and [email protected]:0.95 of 0.574, but testing on real crowded refrigerators exposed missed detections and weak generalization. To address this, a second-generation dataset (V2.1) was prepared for a YOLOv8n-seg model by merging the original FoodLens data with a crowded refrigerator dataset over 30 target classes, giving 9,466 images and 53,349 instances after SHA-256 duplicate removal, and passing the implemented validation checks without reported errors. Training of V2.1 has not been completed, so no V2.1 model performance is reported; comparing V2.1 with V1 on crowded scenes is the immediate next step. The contribution is an integrated, system-oriented workflow rather than a new detection algorithm.},
        keywords = {Food inventory management, YOLOv8, object detection, instance segmentation, smart refrigerator, recipe recommendation, FastAPI.},
        month = {September},
        }