CSOD-24: Construction Site Object Detection Dataset for Safety Monitoring at Construction Site using Deep Learning
Year: 2025
Type: article
Abstract: Monitoring the use of personal protective equipment (PPE) and worker proximity to heavy machinery are two areas where ensuring safety compliance on construction sites continues to be difficult. The lack of dynamic ambient circumstances, comprehensive annotations, and real-time video data in existing datasets restricts their applicability to real-world situations. In order to fill in these gaps, this work presents CSOD-24, a video dataset intended for construction site object detection and safety monitoring. The dataset includes 100 ten-second video clips (16.6 minutes total), covering four major classes: "Dump Truck", "Worker with Helmet", "Worker without Helmet" and "Excavator". The videos were recorded at 10 frames per second (fps) and annotated in .txt, .json, and .xml formats. This dataset supports the development and validation of algorithms for automated safety compliance monitoring, object detection, and tracking in dynamic construction environments. The CSOD-24 dataset address these challenges, enabling a robust foundation for advancing computer vision-based safety monitoring, thereby contributing to reduced workplace hazards and improved operational efficiency.
Institution: Solapur University
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4409824449