High-Resolution Annotated Concrete Column Images for Object Detection

Published: 17 September 2024| Version 1 | DOI: 10.17632/rknjxtg2vn.1
Contributors:
,
,
,

Description

This dataset contains 501 high-resolution images of concrete columns captured from real buildings under construction in Iran. The photos were taken from approximately direct angles, providing clear views of the buildings' exteriors and focusing on the concrete columns located on the outer layer. Each image has been annotated with rectangular bounding boxes to highlight the exterior layer columns, which are at the built stage. The dataset is intended to facilitate detection tasks in machine learning and computer vision applications. The annotations, created using CVAT, are provided in YOLO format. This dataset is particularly valuable for researchers and practitioners working on object detection models, specifically in the fields of structural engineering, construction, and automated inspection systems. Dataset Details: • Total Images: 501 • Image Formats: JPEG and PNG • Annotations: Each image is annotated with a single class representing the boundary box of the concrete columns on the exterior layer of the building. The annotations are provided in YOLO format, with labels as rectangular bounding boxes. • Class: Outer Layer Concrete Column (Built Stage): 8,588 instances • Bounding Box Format: YOLO (x_center, y_center, width, height, normalized by image dimensions)

Files

Institutions

K N Toosi University of Technology Faculty of Electrical Engineering, K N Toosi University of Technology Faculty of Civil Engineering

Categories

Construction, Image Processing, Object Detection, Building

Licence