How Blob Detectors Join Photogrammetry and Computer Vision in 3D Reconstruction: Evolution and Future Developments 

The evolution of three-dimensional reconstruction has been profoundly influenced by the advent of blob detectors, a class of algorithms designed to identify and tie points across scales, perspectives, and illumination conditions. More than a mere technical advancement, blob detectors have acted as a catalyst in joining the rigorous, metric-driven approach of photogrammetry with the adaptive, automation-oriented methodologies of computer vision. This synergy has enabled significant progress, such as robust correspondence even in challenging environments and more scalable workflows for processing increasingly large and heterogeneous datasets.

Their impact is evident in several fields. In cultural heritage, blob detectors have improved digital preservation by ensuring reliable features matching on texture-poor surfaces. In robotics, they have enhanced navigation and spatial awareness, while in environmental monitoring, they have enabled efficient mapping of large and dynamic ecosystems. By reducing manual interventions and accelerating workflows, blob detectors have paved the way for quasi-real-time, data-intensive applications once considered unfeasible.

In the future, the integration of blob detection with artificial intelligence and deep learning opens exciting perspectives. Future developments indicate that hybrid systems capable of learning context-aware feature extraction strategies will dynamically adapt to environmental conditions. Such advancements will not only refine accuracy and efficiency but also broaden the scope of 3D reconstruction, driving it into fields where adaptability and autonomy are essential.

This keynote will explore the trajectory of blob detectors from their algorithmic foundations to their transformative role in joining photogrammetry and computer vision, highlighting both current applications and the promising developments that will define the future of 3D reconstruction.

Mail at: silvio.delpizzo(at)uniparthenope.it

AI4EO Transformation: Foundation Models Changing the Game in Environment Monitoring

Artificial Intelligence is reshaping the landscape of Earth Observation, enabling a new generation of environmental monitoring solutions. This presentation will begin by outlining the fundamental differences between traditional computer vision tasks and the unique characteristics of EO data—such as the diversity of sensors, spectral resolution, and spatiotemporal complexity. 

We will then explore how Planetek Italia integrates EO imagery and machine learning to develop operational analytics and services across a range of applications, including land cover mapping, infrastructure monitoring, and disaster response. These solutions are built on top of years of experience in processing satellite data and are increasingly enhanced by advanced AI techniques.

The second part of the talk will focus on foundation models—a new frontier in AI for EO. We will discuss their potential to generalize across tasks, reduce the need for large labelled datasets, and accelerate model deployment at scale. Drawing from recent initiatives such as the SAI4EO project, we will illustrate how these models can change the game in terms of both performance and usability.

Finally, we will emphasize the strategic importance of collaboration between research institutions and private companies. Such partnerships not only drive scientific advancement but also ensure that cutting-edge technologies translate into real-world, scalable services that support environmental sustainability and policy decision-making.

Mail at: taggio(at)planetek.it