Innovation & Cooperation in Naval Architecture & Marine Engineering Association

Ocean Object Intelligent Perception Dataset

A Joint Industry Project (JIP) for Ocean Object Intelligent Perception Dataset [OOIPD] is proposed. By combining the advantages of multiple technologies, we aim to build an intelligent multi-source sensing dataset (visible images, infrared images, SAR images, point cloud, etc.) driven by multiple tasks such as detection, classification, segmentation, and tracking for marine objects. The goal is to overcome object and obstacle recognition, marine environment reconstruction and other key technologies, achieving autonomous navigation of ships in the marine environment (including polar regions). Application validation will be conducted using test vessels to further achieve objectives such as intelligent sensing, automatic collision avoidance, and trajectory planning in autonomous navigation within specific sea areas.

结合多方技术优势,构建针对海洋目标的检测,分类,分割与跟踪等多任务驱动的海洋目标多源感知数据集(可见光、红外、SAR、点云等),突破船舶自主航行中海洋环境(包括极地环境)目标与障碍物识别,海洋环境重建等关键技术,并依托试验船进行应用验证,进一步实现特定海域自主航行中的智能感知、自动避碰和航线规划等目标。

INITIATING MEMBERS | 发起单位

· Harbin Engineering University | 哈尔滨工程大学

· Wuhan University of Technology | 武汉理工大学

· Dalian Maritime University | 大连海事大学

· Huazhong University of Science and Technology | 华中科技大学

· National Energy Group Shipping Co.,Ltd | 国家能源集团航运有限公司

SCHEDULE | 时间表

· Call for Participates and Introduction Meeting at ICNAME2024 | ICNAME2024会议JIP项目发布会:2024.8.23

· First Meeting at Harbin Engineering Univ. | 第一次会员会,地址哈尔滨工程大学: 2024.8.25

· The framework of Ocean Object Intelligent Perception Dataset is initially established | 初步构建海洋目标智能感知数据集框架,2024.12

· Promoting the construction of Ocean Object Intelligent Perception Dataset for marine intelligent sensing under multiple collaboration. | 在多家协同下构建海洋目标智能感知数据集以用于海洋智能感知,2025.8

· Real ship verification. | 实船验证. 2025.12

CONTACT | 联系方式

JIP Contact:Prof. Su Li | 项目联系人: 苏丽教授 15590897888

suli406@hrbeu.edu.cn

CONSTRUCTION PLAN | 建设方案

哈尔滨工程大学提供海洋目标数据库V1与智能多源感知模型与算法基准库V1,并提出海洋目标数据库构建标准。目前数据库V1版已包括6万张可见光图像,1万张红外图像,1万张遥感图像以及它们相应的标注文件,未来数据库还将加入极地海冰图像,船舶运动视频等。哈尔滨工程大学、武汉理工大学、大连海事大学、华中科技大学以及其他国内外相关单位将共同完善数据集构建标准,增广海洋目标数据库和算法模型库,达到标准发布的条件。

Harbin Engineering University provides the Ocean Object Intelligent Perception Dataset V1 and Ocean Object Intelligent Perception Model and Algorithm Benchmark Library V1, proposing criteria for the construction of marine object dataset. The current dataset version (V1) includes 60,000 visible images, 10,000 infrared images, 10,000 remote sensing images and their corresponding labels. In the future, it will include polar sea ice images, ship motion videos, and more. Harbin Engineering University, Wuhan University of Technology, Dalian Maritime University, Huazhong University of Science and Technology and other domestic and overseas organizations will jointly improve the standards for dataset construction, expanding the dataset and the model library, to meet the conditions for standard publication.

1.jpg

CONSTRUCTION OBJECTIVES | 建设目标

· 项目所构建的海洋目标智能感知数据集经过国内外同行试用,实现广泛推广。

· 成员单位积累和完善自身的海洋目标数据、模型和算法,提升各自在海洋智能感知领域的技术水平和影响力。

· 推动海洋目标智能感知技术在船舶与智能感知交叉领域的进一步发展。

· The Ocean Object Intelligent Perception Dataset developed by the project has undergone iterations and trials by domestic and international experts, achieving widespread promotion.

· Member will accumulate and enhance their own marine data, models, and algorithms, thereby improving their technical capabilities and influence in the field of marine intelligent sensing.

· Promote the further development of intelligent marine object sensing technology at the intersection of the maritime industry and intelligent perception.

INFRASTRUCTURE & FACILITY | 基础与设施 

2.png

海洋目标智能感知国际挑战赛:已成功举办三届,每届有国际、国内约200多支参赛队伍参赛。2020年挑战赛数据类型为可见光图像;2021年红外图像;2022年遥感和SAR图像。

International Challenge on Ocean Objects Intelligent Perception:Successfully held for three editions, each attracting around 200 participating teams from both international and domestic organizations. The competition data for 2020 involved visible images, for 2021 infrared images, and for 2022 remote sensing images and SAR images.

3.jpg

海洋目标数据库V1的部分信息如上。

Some of the information in the Ocean Object Intelligent Perception Dataset V1 is shown above.

FUND & MEMBERSHIP | 资金和会员资格

Harbin Engineering University provides the basic marine object sensing dataset V1 and model algorithm library V1. Each participating unit expands the experimental data through standard protocols, and validates under appropriate experimental conditions. | 哈工程提供基础的海洋目标感知数据库V1版与模型算法库V1版,各参与单位通过标准数据协议扩展智能多源感知数据集,并在适当的实验条件下进行应用验证。

Full membership | 全会员资格:

Full members are required to participate in dataset refinement, labelling, augmentation, etc. Full members get access to the full dataset. Full members are exempt from membership fees. | 全会员需参与数据集的完善、标注、增广等工作,可获得全部数据集使用权。全会员免会员费。

Paid membership | 付费会员资格:

Paid members get access to the full dataset. Membership fees are determined based on final enrolment. | 付费会员获得全部数据集使用权,会员费根据最终报名情况确定。

Limited membership | 有限会员资格:

Limited members receive access to a portion of the dataset, and can use a portion of the dataset for public testing, evaluation, and competition. Limited members are exempt from membership fees. | 有限会员获得部分数据集使用权,可利用部分数据集进行公开的测试、评估、竞赛。有限会员免会员费。