[Paper Review] Extending 3-DoF Metrics to Model User Behaviour Similarity in 6-DoF Immersive Applications

Silvia Rossi, Irene Viola+2 more|arXiv (Cornell University)|Dec 17, 2021

Anomaly Detection Techniques and Applications4 citations

TL;DR

This paper extends 3-DoF user behavior similarity metrics to 6-DoF immersive environments by introducing multi-feature metrics that combine user position and viewing direction, significantly improving similarity detection in volumetric VR. The proposed metric w₇ achieves the highest precision (P > 0.4) across all sequences, outperforming single-feature metrics and enabling robust, real-time clustering of users with similar viewing behavior.

ABSTRACT

Immersive reality technologies, such as Virtual and Augmented Reality, have ushered a new era of user-centric systems, in which every aspect of the coding--delivery--rendering chain is tailored to the interaction of the users. Understanding the actual interactivity and behaviour of the users is still an open challenge and a key step to enabling such a user-centric system. Our main goal is to extend the applicability of existing behavioural methodologies for studying user navigation in the case of 6 Degree-of-Freedom (DoF). Specifically, we first compare the navigation in 6-DoF with its 3-DoF counterpart highlighting the main differences and novelties. Then, we define new metrics aimed at better modelling behavioural similarities between users in a 6-DoF system. We validate and test our solutions on real navigation paths of users interacting with dynamic volumetric media in 6-DoF Virtual Reality conditions. Our results show that metrics that consider both user position and viewing direction better perform in detecting user similarity while navigating in a 6-DoF system. Having easy-to-use but robust metrics that underpin multiple tools and answer the question ``how do we detect if two users look at the same content?" open the gate to new solutions for a user-centric system.

Motivation & Objective

• Address the lack of standardized metrics to model user behavior similarity in 6-DoF immersive VR environments.
• Extend existing 3-DoF behavioral analysis techniques to accommodate full user locomotion and head tracking.
• Develop simple, efficient, and robust similarity metrics suitable for real-time deployment in user-centric multimedia systems.
• Validate the proposed metrics using real navigation trajectories from a 6-DoF VR environment with dynamic volumetric content.
• Enable accurate user clustering for applications such as personalized content delivery, quality of experience optimization, and behavioral profiling.

Proposed method

• Propose seven similarity metrics (w₁ to w₈) that combine user position and viewing direction in 6-DoF space, with w₇ and w₈ being multi-feature combinations.
• Use a ground-truth overlap ratio metric based on common rendered points in the viewport as a benchmark for similarity.
• Apply a clique-based clustering algorithm to group users based on similarity metrics, using the Generalised Dunn’s Index (GDI) and F-measure for evaluation.
• Evaluate metrics using real navigation trajectories from a 6-DoF VR experiment with volumetric point cloud content.
• Compute metrics in real time using standard system data (position and orientation), ensuring low computational overhead (under 0.01 seconds per evaluation).
• Compare performance across frame-based and chunk-based analysis using precision, recall, and F-measure, with ground-truth derived from content overlap.

Experimental results

Research questions

• RQ1 How do 6-DoF user navigation patterns differ from 3-DoF in terms of behavioral complexity and interaction dynamics?
• RQ2 Which combination of user position and viewing direction features best captures behavioral similarity in 6-DoF immersive environments?
• RQ3 Can existing 3-DoF behavioral metrics be extended to 6-DoF systems with minimal computational cost and high accuracy?
• RQ4 How do single-feature versus multi-feature similarity metrics compare in detecting users viewing the same content in 6-DoF VR?
• RQ5 Can the proposed metrics enable reliable, real-time user clustering for user-centric multimedia systems?

Key findings

• Metrics combining both user position and viewing direction (e.g., w₇ and w₈) outperform single-feature metrics in detecting behavioral similarity.
• The w₇ metric achieved the highest precision (P > 0.4) across all sequence time, indicating superior accuracy in identifying users viewing the same content.
• All proposed similarity metrics generated larger clusters than the ground-truth, with w₁, w₅, and w₈ including 80% of the population in relevant clusters.
• The overlap ratio-based ground-truth metric, while accurate, is computationally intensive and impractical for real-time use.
• Despite its simplicity, the single-feature metric w₁ (based only on position) performed comparably to multi-feature metrics, suggesting utility for preliminary analysis.
• The proposed metrics are computationally efficient, with evaluations completed in under 0.01 seconds, making them suitable for real-time deployment in immersive systems.