We are in the midst of the Fourth Industrial Revolution, and so are the bees. The advent of new and more affordable sensors, along with advanced data processing methods, has introduced versatile applications in apidology. Ironically, these technologies are now being employed to better understand the impacts of previous industrial revolutions, including the consequences of changes in land use, climatic factors, and the spread of invasive species facilitated by global trade routes. The shift towards digitized apidology has become crucial in response to the dramatic decline in insect populations worldwide. This dissertation embraces this shift by leveraging and advancing the capabilities of machine learning in computer vision to enhance our understanding and protection of pollinators.The first set of contributions focus on improving video data acquisition around honey bee hives (Apis mellifera) and evaluating devices for quantifying bee mortality for regulatory purposes, thus making plant protection products safer for non-target organisms. A study on the re-identification of bumblebees (Bombus terrestris) demonstrates how future behavioural studies can be conducted without markers, non-invasively and purely visual. Additionally, techniques for the automatic calibration of pollen colours on smartphone images and biodiversity assessment are presented, enabling landscape monitoring through citizen scientists. Finally, we present a groundbreaking honey bee brood monitoring system that can monitor Varroa destructor reproduction in cells, and supports breeding for varroa-resistant traits.