Natalia Piórkowska

This paper compares classical computer vision, deep learning, and generative domain adaptation for cross-session behavioral tracking of Galleria mellonella larvae, finding that classical methods are more robust to domain shifts than deep learning.

This paper reviews AI use in analyzing invertebrate and larval behavior, finding a rise in deep learning applications and calls for standardized reporting.

The paper surveys public behavioral datasets for invertebrate models, evaluates their FAIR compliance, and provides guidance for standardized data sharing and AI benchmarking.