CompetitorFormer: Competitor Transformer for 3D Instance Segmentation

2024-11-21

arXiv.org

AI Summary

In 3D instance segmentation, Transformer-based models suffer from “inter-query competition” caused by redundant learnable queries, leading to degraded convergence and inference stability. Method: This paper proposes a competition-aware mechanism—the first framework to explicitly model and suppress query competition. It introduces a plug-and-play Competition-Guided Module, a Dynamic Competition Weighting strategy, and a Competition-Aware Decoder. Within self-attention, we propose Competition-Aware Attention, coupled with dynamic IoU-weighted query selection and a hierarchical competition suppression loss. Contributions/Results: On ScanNet and S3DIS benchmarks, our method achieves absolute mAP improvements of 2.3–3.8 percentage points. Training convergence accelerates by 30%, while inference accuracy and robustness are simultaneously enhanced—demonstrating superior generalization and efficiency over prior query-based approaches.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Learning to Search
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

Transformer-based methods have become the dominant approach for 3D instance segmentation. These methods predict instance masks via instance queries, ranking them by classification confidence and IoU scores to select the top prediction as the final outcome. However, it has been observed that the current models employ a fixed and higher number of queries than the instances present within a scene. In such instances, multiple queries predict the same instance, yet only a single query is ultimately optimized. The close scores of queries in the lower-level decoders make it challenging for the dominant query to distinguish itself rapidly, which ultimately impairs the model's accuracy and convergence efficiency. This phenomenon is referred to as inter-query competition. To address this challenge, we put forth a series of plug-and-play competition-oriented designs, collectively designated as the CompetitorFormer, with the aim of reducing competition and facilitating a dominant query. Experiments showed that integrating our designs with state-of-the-art frameworks consistently resulted in significant performance improvements in 3D instance segmentation across a range of datasets.

Problem

• 3D Object Recognition
• Transformer-based Instance Querying
• Query Competition

Innovation

• CompetitorFormer
• Transformer-based 3D object recognition
• Efficiency improvement