In a joint effort between engineers from Stanford University (USA) and marine biologists from the Center for Research in Marine Sciences and Limnology (CIMAR) at the University of Costa Rica, FLAIR (Frame Level Alignment and Tracking) was developed. This innovative artificial intelligence tool allows for the automatic and accurate analysis of data on moving animals by tracking this process in real time with drones.
This technology was tested in Santa Elena Bay, within the Santa Elena Bay Marine Management Area of the Guanacaste Conservation Area (ACG), a site of enormous ecological wealth that is home to unique marine ecosystems and endangered species such as the Pacific nurse shark (Ginglymostoma unami). This species is endemic to the Eastern Tropical Pacific and has been little studied.
Technology for marine conservation
The FLAIR system combines image segmentation and recognition models to identify and track sharks in real time, without the need for prior training. This drastically reduces analysis time: a 5-minute video that previously required more than 20 hours of manual annotation can now be processed in less than an hour.
Furthermore, the tool provides key biometric data, such as body length and tail flick frequency, which are essential for understanding the health, behavior, and ecology of these species. Its generalization capabilities also make it useful for studying other marine and terrestrial species, opening up new opportunities for research and conservation.
A Living Laboratory for Science
Santa Elena Bay is a true natural laboratory, where reefs, mangroves, seagrass beds, and a wide diversity of fish, birds, and marine mammals have been documented. This mosaic of habitats not only supports high biodiversity but also provides vital ecosystem services such as coastal protection, sustainable fisheries, and responsible tourism.
Studies like this reinforce the importance of protecting and conserving Santa Elena Bay and the entire Marine Management Area, ensuring that its benefits continue for local communities and the balance of the planet.
Examples demonstrating FLAIR’s ability to segment and track animals in a wide range of videos: https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.70116
Research team:
Chinmay K. Lalgudi (Universidad de Standford)
Mark E. Leone (Universidad de Standford)
Jaden V. Clark (Universidad de Standford)
Sergio Madrigal Mora (CIMAR-Universidad de Costa Rica)
Mario Espinoza (CIMAR-Universidad de Costa Rica)
