Paper Key : IRJ************232
Author: Arfath,Durgaprasad Sl
Date Published: 02 Jan 2025
Abstract
The conversion of 2D blueprints into 3D models is one of the most critical tasks in architecture, engineering, and construction industries. Since the design plans must accurately represent for visualization and implementation, the traditional methods have always been manual, tedious, and prone to errors. With the advancement of computer-aided design (CAD), machine learning, and computer imaging techniques into the automation of these tasks, the conversion is more accurate and efficient. The paper deals with emerging methods to automate the transition from 2D blueprint to its 3D model via deep learning and image processing techniques. The main concern is obtaining a deep understanding of the interpretations of 2D-rendered geometries, various dimensions, and various structural details along with the complex and obscure information it may contain. Methods being developed under convolutional neural networks for feature extraction from the 2D image followed by generative models for reconstruction. A system that operates with semantic segmentation that identifies the entities like walls, doors, and windows. Further, a 3D mesh generation algorithm is used to convert the 2D data into a 3D structure. The methodology utilizes normally applied datasets and benchmarks within architectural and engineering designs for model training and evaluation. Assessment of measurements, computing time, noise-resiliency performances, and scaling due to the noisy real-world nature of blueprint data is done. Some issues with scale and complexity have also been considered, e.g., the requirement for data quality. Future prospects include the addition of augmented reality (AR) visualizations for real-time interactions as well as LiDAR-based sensor data for improved accuracy. This presents great opportunities for rendering extremely possible real-time automation techniques that can be beneficial for shortening design processes and enhancing project outcomes in a number of fields.Key words:coversion of 2D blueprint into 3D model used blender software