Data for An open-source deep learning algorithm for efficient detection of fishes in underwater images from diverse reef ecosystems

FishForYolo.zip comprises images in .jpeg format along with normalized labels in YOLO format and non-normalized labels. This dataset is utilized for both training and testing purposes. The shooting.csv file contains the mapping between videos and their respective shooting instances, which denote all videos captured at a particular location and time. It is used for data analysis. split.yaml delineates the partitioning of data into training, validation, and testing sets. This file is integral for the model to access annotations appropriately during its execution. labels_log.csv contains the annotations in CSV format, supplemented with additional details such as bounding box area or label region. It is instrumental for conducting thorough data analysis. SU_preds.zip contains the predictions of our best model on the test set (in the labels directory). The directory structure mirrors that of FishForYolo.zip. It also contains a file "all_preds.csv" which contains the predictions in CSV format.FishForYolo.zip 包含 .jpeg 格式的图像，以及YOLO（You Only Look Once）格式的归一化标注与非归一化标注。本数据集可用于模型的训练与测试流程。 shooting.csv 文件存储了视频与其对应拍摄实例的映射关系，其中拍摄实例指在特定地点与时间采集的全部视频，该文件用于开展数据分析工作。 split.yaml 文件定义了数据集划分为训练集、验证集与测试集的规则，是模型在运行过程中正确获取标注信息的关键依赖文件。 labels_log.csv 以CSV（Comma-Separated Values）格式存储标注信息，并补充了边界框面积、标注区域等额外细节，可用于开展全面深入的数据分析工作。 SU_preds.zip 包含了我们的最优模型在测试集上的预测结果（存储于labels目录中），该目录的结构与FishForYolo.zip完全一致。此外，该压缩包还包含一个名为"all_preds.csv"的文件，以CSV格式存储所有预测结果。