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

TL;DR

This paper presents RANK-LETR, a Transformer-based line segment detection method that improves ranking accuracy and training efficiency through learnable geometric information, a new proposal method, and a ranking loss.

Contribution

The paper introduces RANK-LETR, which refines line segment confidence scores, proposes a novel proposal method, and stabilizes training with a ranking loss, outperforming existing methods.

Findings

Outperforms other Transformer and CNN-based methods in accuracy.

Requires fewer training epochs than previous Transformer models.

Enhances training stability and prediction quality.

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.