Image Based Recipe Recommendation System

Abstract

The Image Based Recipe Recommendation System presents an intelligent, web-based application that simplifies recipe discovery through image recognition and deep learning. The system enables users to upload or capture images of food, which are then processed using the DenseNet-201 convolutional neural network to extract high-level visual features. These features are compared with pre-encoded representations of a structured recipe dataset in JSON format to retrieve the most relevant recipes. The frontend, built with Django, HTML, CSS, and JavaScript, offers an intuitive interface that displays detailed recipe cards containing dish names, ingredients, cooking instructions, preparation time, and nutritional details. The backend efficiently manages image preprocessing, feature encoding, and similarity computation to ensure rapid and accurate recommendations. The proposed system demonstrates the integration of artificial intelligence, computer vision, and web technologies to bridge the gap between visual food recognition and practical recipe generation.

Introduction

The text describes an AI-based culinary system designed to bridge the gap between visual food identification and recipe retrieval. Traditional recipe search methods rely on text-based queries, which are inconvenient when users only have images of dishes. To solve this, the proposed system enables users to upload food images and automatically identifies the dish and generates corresponding recipes. It addresses challenges such as variations in lighting, background noise, and similarity between food items, while ensuring fast response, scalability, and an easy-to-use interface.

The system is implemented using a web platform built with Django, HTML, CSS, and JavaScript, presenting results in a simple card-based layout that includes dish names, ingredients, and cooking instructions. The core methodology involves image upload or capture, preprocessing, feature extraction using the DenseNet-201, and similarity matching using cosine similarity. The system then retrieves matching recipes from a database and displays them to the user in an organized format.

The workflow is structured through Data Flow Diagrams (DFDs), showing how user-uploaded images are processed by the system, encoded, compared with stored recipe features, and then used to fetch relevant cooking instructions. The Level 0 and Level 1 DFDs highlight the interaction between the user, system, and recipe database.

The model was evaluated using 1,000 food images and achieved an accuracy of 92.4% with an average prediction time of 1.8 seconds. It was compared with other models like VGG16 and ResNet50, where DenseNet-201 performed best in both accuracy and speed.

Conclusion

This System represents a significant stride in the realm of culinary technology, marrying the visual allure of food with the precision of artificial intelligence. Through the amalgamation of state-of-the-art image recognition and natural language processing, this project has endeavored to transform static images into dynamic, interactive cooking experiences. The journey from recognizing ingredients in a mere snapshot to generating coherent and personalized recipes is a testament to the potential of technology in reshaping our culinary adventures. As we reflect on the development process, it becomes evident that the project not only serves as a practical tool for meal planning but also as a source of inspiration for culinary enthusiasts across diverse skill levels. In essence, this system not only encapsulates the capabilities of modern technology but also embodies the spirit of culinary artistry, making every cooking endeavour an exciting and unique experience. It stands as a testament to the evolving intersection of technology and gastronomy, where the boundaries between the virtual and culinary worlds blur, paving the way for a more interactive and delightful cooking future.

References

[1] Y. Tace, et al., “IoT and Machine Learning-Based Smart Irrigation Framework for Sustainable Agriculture,” International Journal of Emerging Technology and Advanced Engineering, vol. 14, no. 2, pp. 52–59, 2025. [2] M. Razak, et al., “Agriculture 5.0 and Explainable AI for Smart Agriculture: A Scoping Review,” IEEE Access, vol. 13, pp. 24412–24427, 2025. [3] H. Wei, et al., “Irrigation with Artificial Intelligence: Problems, Premises, and Promises,” Computers and Electronics in Agriculture, vol. 214, no. 8, pp. 321337, 2024. [4] G. E. O?uztürk, “AI-Driven Irrigation Systems for Sustainable Water Management: Systematic Review and Meta- Analysis,” Elsevier Journal of Agricultural Informatics, vol. 12, no. 1, pp. 17–30, 2025. [5] S. Gupta, “Smart Agriculture Using IoT for Automated Irrigation, Water Conservation, and Crop Monitoring,” IJRASET, vol. 13, no. 4, pp. 401–406, 2025. [6] Z. Zeng, “AI-Driven Smart Agriculture Using Hybrid Transformer–CNN Models for Disease Detection,” IEEE Access, vol. 12, pp. 121344–121356, 2025. [7] P. Rastogi, “Early Disease Detection in Plants Using CNN,” International Journal of Computer Applications, vol. 186, no. 29, pp. 9–15, 2024. [8] A. Y. Ashurov, “Enhancing Plant Disease Detection Through Deep Learning,” Journal of Intelligent Systems, vol. 30, no. 7, pp. 192–199, 2025. [9] X. Liu, “Intelligent and Automatic Irrigation System Based on IoT Using Fuzzy Control Technology,” IEEE Internet of Things Journal, vol. 11, no. 3, pp. 1516–1524, 2025. [10] XenonStack Research Team, “Smart Irrigation: Leveraging Sensor Data and AI for Sustainable Water Management,” Industry White Paper on Precision Agriculture, XenonStack, 2024.

Copyright

Copyright © 2026 Sumeet Kolhe, Prof. M.U. Choudhari . This 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.