Sulawesi ricefish dataset
收藏资源简介:
该数据集由弗劳恩霍夫协会·海因里希赫兹研究所构建,包含苏拉威西米鱼在受控水产环境中的追踪标注视频,分辨率达2448×2048像素/30fps。核心内容包含37帧训练样本和2段测试序列(共202帧),通过半自动标注提供像素级分割掩膜和跨帧身份标签,涵盖鱼群密集、快速转向和频繁遮挡等复杂场景。数据集创建采用Basler专业摄像机多视角采集,模拟家庭水族箱光照条件,旨在为计算机视觉模型开发提供基准,支持通过游泳方向分析等轨迹特征实现鱼类健康监测。
This dataset was developed by the Fraunhofer Heinrich Hertz Institute (HHI). It contains tracked and annotated videos of Sulawesi freshwater fish in a controlled aquaculture environment, with a resolution of 2448 × 2048 pixels at 30 fps. The core content includes 37 training sample frames and two test sequences (totaling 202 frames). Semi-automatic annotation is used to provide pixel-level segmentation masks and inter-frame identity labels, covering complex scenarios such as dense fish schools, rapid turning movements, and frequent occlusions. The dataset was acquired using multi-view recordings with professional Basler cameras, with lighting conditions mimicking those of home aquariums. It aims to serve as a benchmark for computer vision model development, and enables fish health monitoring via trajectory features such as swimming direction analysis.
- 1Video-based Locomotion Analysis for Fish Health Monitoring弗劳恩霍夫协会·海因里希赫兹研究所; 柏林洪堡大学 · 2026年



