Transformer-Based Fusion Outperforms GNNs in Multimodal Robot Vision on ScanNet

This report synthesises findings from 11 peer-reviewed papers addressing the following research question: How does replacing GNN-based fusion with transformer architectures in multimodal robot vision models impact mAP scores for semantic scene understanding on the ScanNet benchmark. In the last few years, the deep learning (DL) computing paradigm has been deemed the Gold Standard in the machine learning (ML) community. Moreover, it has gradually become the most widely used computational approach in the field of ML, thus achieving outstanding results on. 10 claims were extracted from source literature; 10 were independently verified against retrieved documents. An automated multi-reviewer quality assessment produced a score of 8.7/10. This report is a machine-generated literature synthesis and does not constitute original research.

Research goal: How does replacing GNN-based fusion with transformer architectures in multimodal robot vision models impact mAP scores for semantic scene understanding on the ScanNet benchmark?

Autonomous literature synthesis. Automated review score: 8.7/10. Full text and citation available at Assignee Research.