Tom-and-Jerry-VideoGeneration-Dataset
收藏Tom-and-Jerry-VideoGeneration-Dataset
概述
- 任务类别:
- 图像到视频
- 文本到视频
- 语言:
- 英语
- 数据集大小:
- 1K<n<10K
数据集信息
- 场景样本数量: 约6000个
- 视频长度: 6秒
- 帧率: 14帧每秒
- 分辨率: 540x360
数据集格式
. ├── README.md ├── captions.txt ├── videos └── videos.txt
使用示例
python import os from datasets import Dataset, DatasetDict
dataset_dir = lora_dataset/Tom-and-Jerry-VideoGeneration-Dataset captions_file = os.path.join(dataset_dir, captions.txt) videos_file = os.path.join(dataset_dir, videos.txt)
with open(captions_file, r, encoding=utf-8) as f: captions = f.readlines()
with open(videos_file, r, encoding=utf-8) as f: video_paths = f.readlines()
captions = [caption.strip() for caption in captions] video_paths = [video_path.strip() for video_path in video_paths]
assert len(captions) == len(video_paths), f"captions.txt { len(captions)} and {len(video_paths)}videos.txt line not match"
data = { text: captions, video: video_paths }
dataset = Dataset.from_dict(data)
dataset_dict = DatasetDict({ train: dataset }) dataset_dict
许可证
- 许可证类型: Apache-2.0




