Abstract

This dataset is composed of 580 mannually annotated images of shuttle tanker vessels, FPSO and FSO oil rigs. The images were obtained from simple Google search and some of them are 3D models generated from modelling software and used in Gazebo robotics simulations. The images were annotated using Computer Vision Annotation Tool (CVAT), and for each of them there is a .xml Pascal VOC file where the boxes are. The annotated objects were the ship itself and its generic regions, named as bow, mid-ship and stern.

Instructions:

Use the .png images images within the file to train a neural network for object recognition;

Use the .xml Pascal VOC file to obtain the annotated objects (bow, mid-ship and stern) within each image;

Use the VOC_YOLO.py to generate a single .txt file with all annotated boxes within the 580 images to train a YOLOv3 convolutional neural network;

USE de gera_xml.py script to generate or modify the .xml Pascal VOC files;

Datasets

Standard Dataset

Shuttle Tanker, FPSO and FSO image dataset for ship recognition

Abstract

Dataset Files

QUESTIONS?