This research that led to a proof of concept I was developing for a couple of months:

• HouseReader (housereader.com) enables users to understand a residential space from a user-recorded video, automatically generating a report with its layout, household elements, estimated interior cost, and providing various insights.
• It’s an algorithm that combines #AI, #LLMs, #VLMs, #Stitching #ComputerVision (CLIP and SAM) techniques and multiple #Python libraries.
• I’ve documented the journey and some project features: housereader.com/index_project

Published for testing, it’s ready for use just to gather feedback.

Below an example of the report generated by the application after processing a video.

Hope you like it! Any feedback is welcome!