AutoFish
收藏资源简介:
AutoFish是一个用于鱼类细粒度分析的公开数据集,由奥尔堡大学等机构创建。该数据集包含1500张高质量图像,涵盖了454个独特的鱼类样本,每个样本都有实例分割掩码、ID和手动长度测量。数据集的创建过程包括在实验室环境中使用RGB相机采集图像,并通过手动点注释和Segment Anything Model (SAM)生成初始分割掩码,随后进行手动校正。数据集的应用领域包括可持续渔业管理和过度捕捞问题的解决,旨在通过自动化技术提供实时监测和数据分析,以支持渔业管理的透明性和生态保护。
AutoFish is a public dataset for fine-grained fish analysis, developed by institutions including Aalborg University. This dataset contains 1,500 high-quality images covering 454 unique fish specimens, each with instance segmentation masks, IDs, and manual length measurements. The dataset was constructed by collecting images using RGB cameras in a laboratory environment, generating initial segmentation masks via manual point annotations and the Segment Anything Model (SAM), followed by manual correction. Its application fields include sustainable fisheries management and addressing overfishing, aiming to provide real-time monitoring and data analysis through automated technologies to support the transparency of fisheries management and ecological conservation.

- 1AutoFish: Dataset and Benchmark for Fine-grained Analysis of Fish奥尔堡大学视觉分析与感知实验室,丹麦;哥本哈根人工智能先锋中心,丹麦;奥尔堡大学生物与环境科学系,丹麦 · 2025年



