A YOLO Algorithm for Pattern Recognition in Images of Marine Species in Closed Seasons

One of the main problems that arise in the process of extracting marine species during the closed seasons is the indiscriminate loading of marine species that are prohibited because they are in spawning season. These problems are often aggravated by the lack of transparency in inspection process, where inspectors receive bribes when they intervene with vessels on the high seas (Ministerio de la Producción in https://pescayconsumoresponsable.produce.gob.pe/presentacion.html, [1]). In addition to that, (Mar del Perú in https://mardelperu.pe/articulos_wikipesca/reglas-de-juego/, [2]) details how the process of identifying marine species does not have rigorous verification due to not having the appropriate tools and control mechanisms. For those reasons, this paper will show an implementation of a technological solution based on a Artificial Intelligence and mobile application integrated into identification and control processes of marine species in closed seasons by providing an adequate classification of the species detected by those devices. To achieve this, YOLO algorithm has been trained and used in an integrated app. YOLO is an algorithm whose architecture is based on convolutional neural networks. This algorithm, unlike other CNN-based architectures, seeks to perform detection in a single run, which allows it to be an extremely fast alternative by performing two fundamental tasks at the same time: identifying a region of interest and classifying it.