Core Concepts
A coarse-to-fine vision-based approach is proposed to progressively detect and localise Artcodes, enabling connections between physical environments and virtual worlds (the metaverse).
Abstract
The paper addresses the problem of facilitating connections with virtual worlds (the metaverse) within everyday physical environments. It introduces a two-stage vision-based approach, VISIONGUIDE, that leverages Artcodes - interactive surface patterns that can be embedded in everyday objects - to uncover hidden metaverse access points.
The first stage, coarse localisation, generates a heatmap indicating the general presence of Artcodes. The second stage, fine localisation, precisely identifies the exact locations of the Artcodes. This coarse-to-fine approach enables an exploratory interaction process, where people can progressively discover and interact with the metaverse in the physical world.
Detailed experiments are conducted on a new dataset, SWAD, which simulates the deployment of Artcodes in real-world environments. The results demonstrate the effectiveness of the proposed approach in Artcode localisation, highlighting its potential to enable new interaction opportunities between the physical and virtual worlds.
The key factors affecting the localisation performance are identified as lighting conditions, Artcode positions, and deformation. The approach performs well in scenarios with good lighting, proper Artcode placements, and minimal deformations, but faces challenges in environments with poor lighting, extreme camera poses, and cluttered backgrounds.
Stats
The size of most Artcodes in the SWAD dataset is up to 800 × 800.
The SWAD dataset contains 44 images with a total of 117 Artcodes.
Quotes
"As the metaverse continues to proliferate, it is foreseeable that everyday environments will become increasingly saturated with its presence."
"Consequently, uncovering links to these metaverse items will be a crucial first step to interacting with this new augmented world."