fish-vista

Fish-Vista是一个大型的鱼类图像数据集，专注于鱼类视觉特征分析，包含约6万张图片，涵盖1900个不同的物种。它支持物种分类、特征识别和特征分割等多项任务。该数据集包含RGB图像，并提供物种分类、性状识别和性状分割三种配置，其中性状分割任务带有像素级别的标注信息。Fish-Vista的数据来源于博物馆收藏，经过精心的数据处理流程，包括去重、元数据过滤、裁剪和背景移除等步骤。该数据集提供标准化数据操作，并采用CC-BY-NC 4.0许可协议。Fish-Vista is a large-scale fish image dataset dedicated to fish visual feature analysis, containing approximately 60,000 images spanning 1900 distinct species. It supports multiple tasks including species classification, feature recognition and feature segmentation. This dataset provides RGB images and offers three configurations: species classification, trait recognition and trait segmentation, where the trait segmentation task is equipped with pixel-level annotation information. The data of Fish-Vista is sourced from museum collections, and has gone through a meticulous data processing pipeline including deduplication, metadata filtering, cropping and background removal. Standardized data operations are provided for this dataset, which is licensed under CC-BY-NC 4.0.

搜集汇总

背景与挑战

背景概述

Fish-Vista是一个大型的鱼类视觉特征分析数据集，包含约6万张图像，涵盖1900个物种，支持物种分类、性状识别和像素级性状分割任务。数据来源于博物馆收藏，经过严格的数据处理流程，并采用CC-BY-NC 4.0许可协议。