YOLO船舶检测数据集

本数据集面向遥感图像AI研发团队、海事目标检测算法开发者及智能船舶监控系统企业，旨在解决船舶遥感检测中旋转/水平目标标注格式不统一、训练验证集划分零散、多格式转换耗时费力等痛点。数据集对同一批船舶遥感图片同时提供四套标注格式：A套为旋转目标检测YOLO格式（已划分训练/验证集）；B套为旋转目标检测JSON格式（可Labelme编辑）；C套为目标检测VOC格式（支持旋转与水平双任务）；D套为水平目标检测YOLO格式（已划分训练/验证集）。四套格式图片一致，标注信息相互对应。与传统单一格式数据集不同，本数据集使用户可根据任务灵活选择格式，无需重复标注或转换。This dataset is tailored for remote sensing image AI R&D teams, maritime object detection algorithm developers, and intelligent ship monitoring system enterprises, aiming to address the core pain points in ship remote sensing detection, including inconsistent annotation formats for rotated/horizontal objects, scattered training and validation set splits, and time-consuming multi-format conversion work. The dataset provides four sets of annotation formats for the same batch of ship remote sensing images: Format A is the YOLO format for rotated object detection, with training/validation sets already pre-split; Format B is the JSON format for rotated object detection, which is editable via Labelme; Format C is the VOC format for object detection, supporting both rotated and horizontal object detection tasks; Format D is the YOLO format for horizontal object detection, with training/validation sets already pre-split. The images corresponding to the four formats are identical, and their annotation information is fully consistent. Unlike traditional single-format datasets, this dataset enables users to flexibly select the appropriate format based on their specific tasks, eliminating the need for repeated annotation or redundant format conversion.

搜集汇总

背景与挑战

背景概述

YOLO船舶检测数据集是一个面向船舶遥感检测的图像目标检测数据集，属于智慧海洋领域。该数据集针对同一批船舶遥感图片提供了四套对应的标注格式（包括旋转和水平目标检测的YOLO、JSON和VOC格式），旨在解决标注格式不统一、数据划分零散和格式转换繁琐的问题，便于用户根据具体任务灵活选用，无需重复标注或转换。