Multidisciplinary Design and Implementation of a GUI and Application for the automation of an International Fishing Tournament

Artificial intelligence is a subject that every day gains more traction and attention with the media for the accomplishments it can fulfill. Artificial Intelligence branches out in a serial of equal as important but not as known technologies that are just grouped together. The main objective of this project is to create a complete system that be implemented in a real world environment, this case being the live hosting of a tournament, and making it a closed-loop, self-sustainable process in which every manual-to-date process can be automatized, and by making use of convolutional networks, be able to create object detection models and compare several models created by the 3 latest generations of the YOLO architecture so that the best and most efficient model is used, to make predictions on images and videos, in order to automatize the process of fish recognition, fishing practices that normally requires a expert in the field to intervene with. By training and continually evaluating the models created by each of the YOLO algorithm generation training process in each save period, the most efficient model which retained one of the best prediction precision was able to be selected for each of the 2 datasets that were worked on, the best and more efficient models were found with YOLOv5 generation algorithm so the small and nano trained models were selected to be used in the final product.