VDR dataset

该数据集专为船舶吃水读取（VDR）任务构建，包含了用于检测吃水标记、识别吃水刻度以及船舶与水面分割的标注图像。数据集采用VOC格式保存，适用于在多种平台上，如船闸、桥墩、无人机（UAV）和无人水面船（USV）上部署的VDR模型的训练。该数据集规模包含45张图像，涵盖了各种船舶，其中有30张清晰的和15张带污渍的吃水标记。任务涵盖了船舶吃水读取、吃水标记检测、吃水刻度识别以及船舶与水面的分割。

This dataset is specifically constructed for the vessel draft reading (VDR) task, containing annotated images for draft mark detection, draft scale recognition, and vessel-water surface segmentation. It is stored in the VOC format and is applicable for training VDR models deployed on various platforms including lock gates, bridge piers, unmanned aerial vehicles (UAVs), and unmanned surface vehicles (USVs). The dataset comprises 45 images covering various vessels, among which 30 have clear draft marks and 15 have draft marks with stains. The covered tasks include vessel draft reading, draft mark detection, draft scale recognition, and vessel-water surface segmentation.

搜集汇总

背景与挑战

背景概述

该数据集名为MTL-VDR，专注于船舶吃水读数任务，包含吃水标尺检测、吃水刻度识别以及船舶与水的分割。数据集提供了百度网盘和Google Drive的下载链接，便于用户获取和使用。