Improving Transformer Based Line Segment Detection with Matched Predicting and Re-ranking

2025-02-25

AI Summary

Existing Transformer-based line segment detectors suffer from two key limitations: (1) high-accuracy predictions are erroneously suppressed due to underestimated confidence scores; and (2) reliance on bipartite matching incurs high computational overhead and slow convergence. To address these bottlenecks, we propose RANK-LETR, featuring three core innovations: (1) a geometry-aware posterior re-ranking mechanism that recalibrates confidence scores based on line segment length, orientation, and positional consistency; (2) a direct line proposal strategy leveraging centroid-proximal feature points, eliminating redundant bipartite matching; and (3) a differentiable line ranking loss to enhance ranking robustness and generalization. Experiments demonstrate that RANK-LETR achieves superior accuracy over state-of-the-art CNN- and Transformer-based methods, reduces training epochs by ~40%, converges significantly faster, and exhibits enhanced robustness to input noise.

Technology Category

• Computer Vision: Segmentation
• Search and Optimization: Learning to Search
• Machine Learning: Learning Preferences or Rankings

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: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Classical Transformer-based line segment detection methods have delivered impressive results. However, we observe that some accurately detected line segments are assigned low confidence scores during prediction, causing them to be ranked lower and potentially suppressed. Additionally, these models often require prolonged training periods to achieve strong performance, largely due to the necessity of bipartite matching. In this paper, we introduce RANK-LETR, a novel Transformer-based line segment detection method. Our approach leverages learnable geometric information to refine the ranking of predicted line segments by enhancing the confidence scores of high-quality predictions in a posterior verification step. We also propose a new line segment proposal method, wherein the feature point nearest to the centroid of the line segment directly predicts the location, significantly improving training efficiency and stability. Moreover, we introduce a line segment ranking loss to stabilize rankings during training, thereby enhancing the generalization capability of the model. Experimental results demonstrate that our method outperforms other Transformer-based and CNN-based approaches in prediction accuracy while requiring fewer training epochs than previous Transformer-based models.

Problem

• Improving line segment detection accuracy
• Reducing training time complexity
• Enhancing model generalization capability

Innovation

• Learnable geometric information refinement
• New line segment proposal method
• Line segment ranking loss introduction