This is the authors' abstract. We don't add key points for this paper.
Pavement crack detection plays a crucial role in maintaining road safety and infrastructure integrity. Existing methods have designed a variety of complex networks to ac-complish this task. However, training a network from scratch is often time-consuming, and the intricate designs can obscure the fundamental nature of the task, making it challenging for researchers to approach. To address this problem, this paper presents an innovative universal framework for crack detection. Instead of designing complex networks from scratch, we directly leverage the existing pre-trained segmentation methods through the use of visual prompts. To let pre-trained segmentation methods better transfer to the crack detection task, we design the visual crack prompt (VCP) mechanism, which encourages the original segmentation model to pay closer attention to high-frequency features, thereby enhancing its ability to detect and segment specific features effectively. Experimental results demonstrate the framework's superior performance across all five datasets used in this paper.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Kuang et al. (2024) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: