视频数据

数据集产品为网剧、短剧、电视剧等影视剧集视频剪辑片段，仅供大模型算法训练场景使用。上述场景可服务时长超过10万小时，包含战争、情感、都市、古装等多种题材。视频格式可提供mp4、mov等格式。分辨率可按需求提供4K清晰度、高清或标清分辨率。比特率可按需求制作提供。所有视频带声音，音频采样率22Khz以上。视频不包含水印、台标、字幕等文字信息。

This dataset product consists of video clip snippets from film and television works including online dramas, short dramas, TV series, and similar content, and is exclusively for use in large language model algorithm training scenarios. The total available duration of this dataset exceeds 100,000 hours, covering multiple genres such as war, romance, urban, historical costume, and others. Multiple video formats are offered, such as MP4, MOV, etc. Resolution options, including 4K, high-definition (HD), and standard-definition (SD), can be provided according to user requirements. The bitrate can also be customized and supplied as requested. All videos are equipped with audio that has a sampling rate of above 22 kHz. No text information such as watermarks, TV station logos, or subtitles is present in the videos.

搜集汇总

背景与挑战

背景概述

该视频数据集包含网剧、短剧等影视剪辑片段，适用于大模型训练，提供超10万小时多种题材内容，支持多种格式、分辨率和比特率，且为无字幕无台标的高质量音视频素材。