The AFFINE project aims to synergistically use geometric and visual data acquired from different sensors to obtain, for the same amount of resources used for acquisition, extremely dense and geometrically much more detailed models, substantially improving the competitiveness of the digitization system, and making it ready for the demands of modern industrial and commercial applications.
From the perspective of innovation, the project aims to:
- research intelligent methodologies, mainly based on deep learning approaches, to densify individual geometric scans by exploiting the information in panoramic images;
- research scalable out-of-core solutions, to effectively integrate densified data from individual scans to obtain a single dense 3D point cloud representative of the entire area or building acquired;
- combine geometric, visual, and odometric data to improve the mutual alignment of 3D data, imagery, and the location itself within the generated model;
- use high-resolution images with densified geometric information and of the densified 3D model on both data sharing and information cataloging and information extraction and modeling platforms to improve existing products and processes and to introduce new features.
