CARE-ON

Can We Rely on our Network?
CARE-ON (Can we REly on Our Network?) is a research project conducted and funded by the Institute of Geosciences and Earth Resources of the Italian National Research Council (IGG – CNR). Its main goal is to investigate the variability related to human error in the manual counting of geometric elements on visual media.
Specifically, the case study chosen concerns the manual counting of boulders and craters from images of extraterrestrial surfaces, a context of fundamental importance for planetary science, where unfortunately it is not possible to conduct on-site observations, except in rare cases. Although machine learning algorithms have advanced in automating these tasks, human capabilities in image identification and interpretation remain essential for their training, accuracy control, and result validation. Manual methods, on the other hand, inevitably introduce a component of variability, due to different individual sensitivities, fatigue, and the observer’s competence/experience.
CARE-ON aims to explore and mitigate such uncertainties by assessing the impact of human factors in this process and developing guidelines that maximize reliability and accuracy in identification. The project also tests for the first time the potential contribution of non-professional volunteer operators in the identification and counting procedures, conducting comparative analyses of the reliability of results obtained by expert researchers, students, and amateurs.
CARE-ON openly shares its results, encouraging the adoption of protocols and guidelines, and promoting an inclusive approach to planetary scientific research.
Project’s goals:
- Defining best practices for the visual counting of boulders and craters
- Estimating operator-related factors affecting data quality and reliability
- Testing participatory approaches to empower citizen scientists
Start date: October 2024
Duration: 18 months
CNR scientific contact: Dr. Sandro Rossato, CNR-IGG (sandro.rossato@igg.cnr.it)
Website: https://care-on.cnr.it/