# Improving Social Awareness Through DANTE: A Deep Affinity Network for Clustering Conversational Interactants

## TL;DR

This paper introduces DANTE, a deep learning-based method for detecting conversational groups in social scenes, improving accuracy over previous methods and demonstrating practical use in human-robot interaction.

## Contribution

The paper presents a novel Deep Affinity Network (DANTE) that predicts pairwise affinities for group detection, combining deep learning with clustering for improved social awareness.

## Key findings

- DANTE outperforms prior methods on established benchmarks.
- The approach effectively detects both small and large conversational groups.
- Practical application demonstrated in human-robot interaction scenarios.

## Abstract

We propose a data-driven approach to detect conversational groups by identifying spatial arrangements typical of these focused social encounters. Our approach uses a novel Deep Affinity Network (DANTE) to predict the likelihood that two individuals in a scene are part of the same conversational group, considering their social context. The predicted pair-wise affinities are then used in a graph clustering framework to identify both small (e.g., dyads) and large groups. The results from our evaluation on multiple, established benchmarks suggest that combining powerful deep learning methods with classical clustering techniques can improve the detection of conversational groups in comparison to prior approaches. Finally, we demonstrate the practicality of our approach in a human-robot interaction scenario. Our efforts show that our work advances group detection not only in theory, but also in practice.