1. Reconstruction of a indoor room from multiple view images
- Author
-
Cid Munuera, Guillermo, Universitat Politècnica de Catalunya. Departament de Ciències de la Computació, Universitat de Barcelona, Universitat Rovira i Virgili, Sànchez-Marrè, Miquel, and Martínez Moré, Gregorio
- Subjects
Computer Vision ,Transferència de colors ,Visió per ordinador ,Detecció i Segmentació ,Camera calibration ,Calibració de Càmera ,Three-dimensional imaging ,Instance segmentation ,Object classification ,Informàtica::Intel·ligència artificial [Àrees temàtiques de la UPC] ,Computer vision ,Visió per computadora ,Classificació d'objectes ,Color matching ,Imatgeria tridimensional - Abstract
The aim of this works is developing a software system that uses machine learning and computer vision to reconstruct a real indoor environment into a 3D model for use in interior design. The project involves assembling a robust computer vision system, which will require the use of deep learning and image processing techniques. The software will take a set of photos of a room as input, apply camera calibration algorithms to find the correct camera parameters, build a reconstruction of the room in 3D, apply instance segmentation to locate different elements in the room, generate 3D models of these objects, and finally, obtain the correct material and texture for each element of the 3D reconstruction. The project requires dividing the work into different sub-jobs, with each requiring its own research to achieve the most satisfactory solution, taking into account aspects such as efficiency, time, and computer resources.
- Published
- 2023