Abstract
Riprap rocks and large-sized aggregates have been used extensively in geotechnical and hydraulic engineering applications to serve as a key component for erosion/sediment control and scour protection. Toward sustainable and reliable use of riprap rocks, an efficient and accurate method to characterize the size and shape properties is deemed necessary. Current state-of-the-practice methods mostly assess riprap properties with labor-intensive and time-consuming inspection routines that involve manual size and weight measurements. Recent advances in the field of computer vision have been leveraged in this paper to apply deep learning for the development of an innovative image analysis tool for quantitative and efficient riprap characterization. Based on the single-aggregate and stockpile-aggregates studies conducted on this topic, this paper introduces a newly developed computer program, named I-RIPRAP, for the advanced characterization of aggregate size and shape from images of riprap stockpile(s). The essential features and software workflow, as well as the function and mechanism of each module, are described and discussed in detail. In addition to the deep learning methods for segmentation, I-RIPRAP also improves the morphological analyses with a volume/weight estimation module and a riprap category reference module to generate useful results that can facilitate quality assurance/quality control tasks. The I-RIPRAP software is envisioned to serve as an efficient and innovative tool for field and in-place evaluations of riprap and large-sized aggregates.
Get full access to this article
View all access options for this article.