Back to Projects

Espai

An AI-powered interior design assistant utilizing diffusion models for real estate staging and space redesign.

2023Co-Creatorwith Aliu Mujib

Overview

Espai (pronounced ess-pie) is a tool built to simplify interior design and furniture discovery using computer vision and generative AI. It began as a personal utility to visualize how furniture from Pinterest or styled room images would look in a personal space, later evolving into a fully featured product bridging the gap between aesthetic inspiration and practical implementation.

The Challenge

The primary goal was to automate the tedious process of identifying furniture in images and finding purchase options. It required solving complex problems like advanced image segmentation, object detection, and image-to-image AI workflows to professionally restyle or stage rooms digitally.

Key Features

  • AI-powered furniture identification
  • Computer vision for object detection & segmentation
  • Generative AI for room restyling
  • Reverse search for purchase links
  • Digital room staging & placement

The Solution & Workflow

1. Identification

Integrated Google's OWL-ViT with OpenCV to allow users to select individual furniture items and get real-time shopping links.

2. Restyling

Implemented Stable Diffusion XL to transform entire rooms into different interior aesthetics (e.g., mid-century to minimalist).

3. Staging

Used edge detection and inpainting workflows to mask objects and regenerate background content cleanly.

Outcomes

Espai validated core ideas around practical visual AI and resonated with early testers. While it was eventually sunsetted to focus on newer ventures, it served as a robust technical foundation for generative product work.

Tech Stack

GolangFlutterPythonOpenCVSDXLSegment AnythingLanceDB