Automatic identification of solar and extra-solar moving and transiting objects in astronomical images with Artificial Intelligence

The rapid growth of photometric data from space- and ground-based surveys made crucial the use of Artificial Intelligence (AI) in the search for moving objects within and beyond the Solar System. This doctoral thesis addresses two critical challenges in modern astrophysics: optimizing Deep Learning (DL) architectures for exoplanet detection and providing the observational constraints to characterize the dust environment of long-period comets in support of the ESA/Comet Interceptor mission. The core thesis of the exoplanetary research is that the high complexity and overparameterization of state-of-the-art Convolutional Neural Networks (CNNs) are redundant for transit detection. Since planetary transits are essentially low-dimensional features, I demonstrate that minimizing architectural depth, width, and resolution mitigates overfitting and enhances generalization. Through a methodology involving compact CNNs and dimensionality reduction techniques, I designed models with cross-mission compatibility, specifically demonstrated using data from both NASA's Kepler and the Transiting Exoplanet Survey Satellite (TESS). Results show that simplified architectures perform at least on par with complex ones, providing a robust framework for future surveys like ESA's PLAnetary Transits and Oscillations of stars (PLATO). In parallel, in support to the selection of the target comet for the ESA/Comet Interceptor mission, I served as principal observer for the acquisition and analysis of photometric data of long-period comets using the Copernico Telescope (Asiago, Italy) and the Telescopio Nazionale Galileo (TNG, Canary Islands, Spain). I reduced these data by using a pipeline written in Interactive Data Language to derive $Af\rho$ curves. These curves were subsequently used to impose constraints on the study aimed at characterizing the composition and activity of the dust tails of long-period comets located more than 4 astronomical units from the Sun.