This Thesis aims to solve the maritime traffic monitoring problem through vessel detection and classification. Maritime surveillance has a significant role for safety, traffic regulation, and port management. The main goal of this work is to design a vision-based system that where ships can be detected in complex maritime scenes and their type can be classified based on visual information. The proposed system is based on state-of-the-art object detection models from the YOLO (You Only Look Once) family, selected for their balance between accuracy and real-time performance. Different model variants are trained and evaluated in order to analyze their behavior on maritime scenarios.