AI-Powered Multimodal Culinary Canvas
Click the Poster to View Full Screen, Right click to save image
Brendan Hannon
College:
The Dorothy and George Hennings College of Science, Mathematics, and Technology
Major:
Computer Science
Faculty Research Advisor(s):
Yulia Kumar
Abstract:
In the contemporary culinary landscape, the quest for personalized recipe recommendations that resonate with individual preferences, dietary requirements, and ecological consciousness is more pronounced than ever. Addressing this need, the "AI-Powered Multimodal Culinary Canvas" emerges as a sophisticated and streamlined recipe recommendation system. It employs a blend of cosine similarity and Term Frequency-Inverse Document Frequency (TF-IDF), further enhanced by integrating an advanced AI cooking Assistant to provide customized recipe suggestions based on available ingredients.
The core of the system is in its multimodality. The incorporation of OpenAI's DALL-E 3 enriches the system with sophisticated image generation; cutting-edge ChatGPT-4-Vision provides the ability to recognize the images of available food, OpenAI's Whisper technology introduces an intuitive voice command feature, elevating the ease of interaction, especially during the culinary process. This system's unique ability to accept voice and image inputs significantly augments its accessibility and user engagement. The AI-driven image generation enhances the visual appeal of recipe suggestions and aids in ingredient identification and recipe visualization. The integration of auditory, textual, and visual modalities, complemented by a responsive feedback mechanism, signifies a substantial leap forward in recipe recommendation systems. It also exemplifies the usage of AI in various application domains, validating the capabilities of the latest AI models.