In our modern world, objects play a central role in shaping our lives, with platforms like Amazon offering unprecedented access. The main idea is to investigate how vastly is possible to define a space with a textual (how do we do architecture with texts and definitions) and actually possibility to realize that space using the real data of the Amazon.
Although image to text models are very powerful to speculate and improve design approaches in many ways, this project aims to invest in the definition of the each object in a space.
A potential solution leverages ML algorithms to bridge this gap, using Language Model (LLM) outputs to create room collages from text inputs and Amazon images.
During the scope of this semester, A chair dataset was scraped from Ikea and Amazon with the limited focus on diverse chair images.
Utilizing Language Model (LLM) outputs, such as those generated by the ChatGPT API, to describe objects in a comprehensive manner could give us diverse applicable object oriented spaces. These descriptions encompass various attributes. including materials, colors, categories, and spatial relations. By leveraging the vast repository of object images available on Amazon & Ikea we can associate these textual descriptions with corresponding visual representations.
