Guangdong Science and Technology Department

Concrete surface crack detection plays a crucial role in infrastructure maintenance and safety. Deep learning-based methods have shown great potential in this task. However, under real-world conditions such as poor image quality, environmental interference, and complex crack patterns, existing models still face challenges in detecting fine cracks and often rely on large training parameters, limiting their practicality in complex environments. To address these issues, this paper proposes a crack detection model based on adaptive feature quantization, which primarily consists of a maximum soft pooling module, an adaptive crack feature quantization module, and a trainable crack post-processing module. Specifically, the maximum soft pooling module improves the continuity and integrity of detected cracks. The adaptive crack feature quantization module enhances the contrast between cracks and background features and strengthens the model’s focus on critical regions through spatial feature fusion. The trainable crack post-processing module incorporates edge-guided post-processing algorithms to correct false predictions and refine segmentation results. Experiments conducted on the Crack500 Road Crack Dataset show that, the proposed model achieves notable improvements in detection accuracy and efficiency, with an average F1-score improvement of 2.81% and a precision gain of 2.20% over the baseline methods. In addition, the model significantly reduces computational cost, achieving a 78.5–88.7% reduction in parameter size and up to 96.8% improvement in inference speed, making it more efficient and deployable for real-world crack detection applications.

BackgroundThe high incidence, seasonal pattern and frequent outbreaks of hand, foot and mouth disease (HFMD) represent a threat for billions of children around the world. Detecting pre-outbreak signals of HFMD facilitates the timely implementation of appropriate control measures. However, real-time prediction of HFMD outbreaks is usually challenging because of its complexity intertwining both biological systems and social systems.ResultsBy mining the dynamical information from city networks and horizontal high-dimensional data, we developed the landscape dynamic network marker (L-DNM) method to detect pre-outbreak signals prior to the catastrophic transition into HFMD outbreaks. In addition, we set up multi-level early warnings to achieve the purpose of distinguishing the outbreak scale. Specifically, we collected the historical information of clinic visits caused by HFMD infection between years 2009 and 2018 respectively from public records of Tokyo, Hokkaido, and Osaka, Japan. When applied to the city networks we modelled, our method successfully identified pre-outbreak signals in an average 5 weeks ahead of the HFMD outbreak. Moreover, from the performance comparisons with other methods, it is seen that the L-DNM based system performs better when given only the records of clinic visits.ConclusionsThe study on the dynamical changes of clinic visits in local district networks reveals the dynamic or landscapes of HFMD spread at the network level. Moreover, the results of this study can be used as quantitative references for disease control during the HFMD outbreak seasons.

With the development of data backup technology, there have been a variety of current disaster recovery system model. This paper has chosen three data backup and recovery system to introduce and analyze according to the current basic data backup and recovery system model, designed to make the reader have a clear and thorough understanding of the current disaster recovery system that has already established.

Increasing quantitative data generated from transcriptomics and proteomics require integrative strategies for analysis. Here, we present an R package, clusterProfiler that automates the process of biological-term classification and the enrichment analysis of gene clusters. The analysis module and visualization module were combined into a reusable workflow. Currently, clusterProfiler supports three species, including humans, mice, and yeast. Methods provided in this package can be easily extended to other species and ontologies. The clusterProfiler package is released under Artistic-2.0 License within Bioconductor project. The source code and vignette are freely available at http://bioconductor.org/packages/release/bioc/html/clusterProfiler.html.