Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{171672,
        author = {Mohammed Shezan and Dr. Nasreen Fathima and Niha Rehaman and Preksha Jain M and Shishira N},
        title = {Conversion of 2D Blueprints into 3D Models},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {8},
        pages = {566-571},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=171672},
        abstract = {We propose a method converting to 2D blueprints to 3D models the deformable object categories from raw single- view images, entirely without external supervision. Our approach leverages an autoencoder framework that decomposes each input image into four fundamental components: depth, color correction, viewpoint, and illumination. This decomposition is achieved without explicit labels. We exploit the fact of converting object appearance, the underlying structure often remains symmetric, which can be used to guide the disentanglement process. To handle objects that may exhibit partial symmetry, we introduce a learned symmetry probability map, which is integrated into the model and learned end-to- end alongside the other components. Our method is capable of accurately recovering the 3D shape of various deformable objects, such as human faces, cat faces, and cars, from single-view images, without relying on any supervision or prior shape models. In experimental evaluations, we demonstrate that our unsupervised method significantly outperforms a supervised approach that relies on 2D image correspondences, achieving superior accuracy in 3D shape reconstruction. This work presents a promising step toward unsupervised 3D object learning, with potential applications in computer vision and graphics.},
        keywords = {Opencv, depth processing, Midas, image processing, pytorch.},
        month = {January},
        }