DUO dataset

The annotations folder contains tags in. json format for training, testing, and verification sets.The labels folder contains labels in. txt format for training, testing and verification sets, which is convenient for yolo model.The images folder contains pictures of the training, testing and verification sets of the original DUO data set.The Gamma_clahe_image folder contains the pictures of the training, testing and verification set of the data set after the image enhancement of the Gamma_clahe algorithm.There are 6671 pictures in the training set, 1111 pictures in the test set and 1111 pictures in the verification set. The total number of targets in the data set is 74,515, of which the numbers of sea cucumber, echinacorus japonicus, scallop and starfish are 7,887, 50,156, 1924 and 14,548 respectively.

标注文件夹内含用于训练集、测试集与验证集的JSON格式标注标签。标签文件夹存储了适配YOLO模型的训练、测试及验证集TXT格式标签。图像文件夹存放了原始DUO数据集的训练、测试与验证集图像。Gamma_clahe_image文件夹则存储了经Gamma-CLAHE算法完成图像增强后的数据集的训练、测试与验证集图像。本数据集训练集共计6671张图像，测试集与验证集各含1111张图像，数据集总目标实例数为74515，其中海参、光棘球海胆（Echinacorus japonicus）、扇贝与海星的目标数量分别为7887、50156、1924与14548。

搜集汇总

背景与挑战

背景概述

DUO数据集是一个用于水下目标检测的数据集，包含训练集（6671张图片）、测试集（1111张图片）和验证集（1111张图片），总计74,515个目标。数据集经过Gamma_clahe算法增强，适用于YOLO模型，并包含海参、扇贝等多种海洋生物目标。