WLD 动物目标检测数据集

WLD 全称 WildLife Documentary，是一个用于动物目标检测的数据集。该数据集包含从 YouTube 下载的 15 部纪录片，其时长从 9 分钟到长达 50 分钟不等，并且总帧数超过 747,000 。该数据集标注了 65 个类别的 4,000 多个对象的轨迹。该数据集的一个特征是所有视频都附有由 YouTube 语音自动生成的字幕。

WLD, short for Wildlife Documentary, is a dataset dedicated to animal object detection. It contains 15 documentaries downloaded from YouTube, with durations ranging from 9 minutes to as long as 50 minutes, and the total number of frames exceeds 747,000. This dataset has annotated trajectories of over 4,000 objects across 65 categories. A distinctive feature of this dataset is that all videos are accompanied by automatically generated subtitles via YouTube's voice recognition service.

搜集汇总

背景与挑战

背景概述

WLD动物目标检测数据集（WildLife Documentary）是一个基于15部YouTube纪录片视频构建的数据集，总帧数超过74.7万，标注了65个类别的4,000多个对象轨迹，并包含自动生成的字幕。