Deep Matching for structure and motion

  1. Home
  2. Research
  3. Projects
  4. Deep Matching for structure and motion
Starting date
November 15, 2023
Duration (months)
14
Departments
Computer Science
Managers or local contacts
Castellani Umberto

Study and developing a new method for conjugate points computation in a structure and motion framework. A data-driven approach is considered focusing on the recent advances on deep learning research with special emphasis on new neural networks based on graph structure and transformer architectures. Sparce matching on wide multiple baseline scenarios is addressed for challenging situations derived by changing of light conditions and strong variations of camera poses

Sponsors:

3DFLOW SRL
Funds: assigned and managed by the department

Project participants

Umberto Castellani
Full Professor
Research areas involved in the project
Intelligenza Artificiale
Machine learning  (DI)
Ingegneria del Software e Verifica Formale
Machine learning  (DI)

Activities

Research areas

Research groups

PhD Programmes

Research facilities

Share