• Two Artificial Intelligence algorithms capable of recognizing an individual and the species to which it belongs were created, said Master Jorge Luis Compean
• The CITEDI graduate stressed that this is intended to provide better control for the recognition and conservation of this species
• With this system formed a database with the different species that nest in Mexico
Through the use of Artificial Intelligence researchers from the National Polytechnic Institute (IPN) developed a system of recognition and classification of sea turtles, unique in its kind, which seeks to provide better control for the recognition and conservation of this species, whose hunting and exploitation have placed it in a position of high vulnerability.
The development of the digital tool was in charge of Jorge Luis Compean Aguirre, graduate of the Center for Research and Development of Digital Technology (CITEDI), who took up the experience of researchers, caregivers and people related to the turtle fields, who, from experience, know many biometric features, such as color, size, shape of the shell and its shields, the facial marks and the scales of the head, which formed a database with the different species that nest in Mexico.
"The classic technique to identify the species is based on the visual recognition of certain biometric features, and in the case of individuals, a numbered plate is placed on the fin, but it is an invasive method, since it is necessary to pierce the limb of the turtle, which can tear and injure the animal if it falls or is stuck, " highlighted the Master of Science in Digital Systems With the advice of professors Ciro Andres Martinez Garcia Moreno and Alejandro Alvaro Ramirez Acosta, from CITEDI, Jorge Compean developed an Artificial Intelligence algorithm to extract characteristics directly related to shape, texture and color, in order to concatenate it into an image and train a convolutional neural network developed by himself, which functions as the neurons of the primary visual cortex of the brain. The polytechnic team, in which participated Dr. Mireya Sarai Garcia Vazquez, from CITEDI and Dr. Miguel Ángel Reyes Lopez, from the Center for Genomic Biotechnology (CBG), also developed another database with photographs of 30 individuals, acquired in Colola, Michoacan, from whom 15 photographs were obtained from each (five on the left side, five on the right side and five with top view of the head), to develop a recognition algorithm and train another artificial neural network.
“This second algorithm can identify the facial marks of the turtles' faces, and other points of interest such as shape and texture, which would be like their fingerprint, to know which individual it is. It is the same principle of the algorithm used by the social network that suggests labeling a person who has already been recognized by the face, but applied to this species”, said the also Engineer in Electronics.
According to Compean Aguirre, the system developed is able to correctly identify and classify both the individual, as well as the species to which it belongs, with 96 percent accuracy, so the next stage would be the implementation of these architectures and algorithms, in a mobile device, to use it on the beaches that are intended for the preservation and conservation of the species.
