Publication Date: 2024/02/03
Abstract: In the construction, manufacturing, and various industrial sectors, the routine task of manually counting steel rods is known to be laborious, time- consuming, and error-prone. This paper focuses on addressing the challenge of object detection and accurate counting of steel rods—a task of critical importance across various computer vision applications. The approach taken here excels in swiftly, accurately, and robustly counting steel rods under diverse conditions encountered at construction sites. The workflow commences with the conversion of video content into image format, a preliminary step that streamlines subsequent processes, including annotation and augmentation. For the automated counting of rods, a diverse range of models were employed, designed to process both images and videos. These models not only provide precise counts of the rods but also furnish valuable insights into rod diameter, enhancing the depth of information available. In this article, a comprehensive comparison of model accuracies can be found, which is a crucial step in identifying the best-performing model for this task. The model proposed here offers a transformative solution by eliminating the need for manual counting efforts, effectively mitigating the potential for human errors that plague traditional counting methods. Powered by advanced image processing techniques, this system not only accelerates the counting process but also substantially enhances accuracy. Consequently, it introduces substantial time and cost savings for construction projects.
Keywords: Counting Steel Rods, Artificial Intelligence, Image Preprocessing, Advanced Models.
DOI: https://doi.org/10.5281/zenodo.10613245
PDF: https://ijirst.demo4.arinfotech.co/assets/upload/files/IJISRT24JAN1201.pdf
REFERENCES