Line Segment Detection Using Transformers without Edges

6 Jan 2021  ·  Yifan Xu, Weijian Xu, David Cheung, Zhuowen Tu ·

In this paper, we present a joint end-to-end line segment detection algorithm using Transformers that is post-processing and heuristics-guided intermediate processing (edge/junction/region detection) free. Our method, named LinE segment TRansformers (LETR), takes advantages of having integrated tokenized queries, a self-attention mechanism, and an encoding-decoding strategy within Transformers by skipping standard heuristic designs for the edge element detection and perceptual grouping processes... We equip Transformers with a multi-scale encoder/decoder strategy to perform fine-grained line segment detection under a direct endpoint distance loss. This loss term is particularly suitable for detecting geometric structures such as line segments that are not conveniently represented by the standard bounding box representations. The Transformers learn to gradually refine line segments through layers of self-attention. In our experiments, we show state-of-the-art results on Wireframe and YorkUrban benchmarks. read more

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Line Segment Detection wireframe dataset LETR F1 score 0.833 # 1
sAP10 65.2 # 5
sAP15 67.7 # 3
Line Segment Detection York Urban Dataset LETR F1 score 0.669 # 1
sAP10 29.4 # 3
sAP15 31.7 # 2