anime_person_detection

Dataset for anime person detection.

| Dataset | Train | Test | Validate | Description |
|-------------|-------|------|----------|---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| v1.1 | 9255 | 460 | 877 | Annotated on the Roboflow platform, including labeled data for various types of anime images (e.g. illustrations, comics). The dataset has also undergone data augmentation techniques to enhance its diversity and quality. |
| raw | 3085 | 460 | 877 | The same as `v1.1` dataset, without any preprocess and data augmentation. Suitable for directly upload to Roboflow platform. |
| AniDet3.v3i | 16124 | 944 | 1709 | Third-party dataset, source: https://universe.roboflow.com/university-of-michigan-ann-arbor/anidet3-ai42v/dataset/3 . The dataset only contains images from anime series. This means the models directly trained on it will not perform well on illustrations and comics. |

The best practice is to combine the `AniDet3.v3i` dataset with the `v1.1` dataset for training. We provide an [online demo](https://huggingface.co/spaces/deepghs/anime_object_detection).

动漫人物检测数据集。

| 数据集 | 训练集 | 测试集 | 验证集 | 描述 |
|---------|--------|--------|--------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| v1.1 | 9255 | 460 | 877 | 在Roboflow平台上完成标注，涵盖各类动漫图像（如插画、漫画）的标注数据。本数据集已通过数据增强技术提升其多样性与数据质量。 |
| raw | 3085 | 460 | 877 | 与`v1.1`数据集完全一致，未经过任何预处理与数据增强操作，适合直接上传至Roboflow平台。 |
| AniDet3.v3i | 16124 | 944 | 1709 | 第三方数据集，来源：https://universe.roboflow.com/university-of-michigan-ann-arbor/anidet3-ai42v/dataset/3。该数据集仅包含动漫剧集类图像，因此基于其直接训练的模型在插画与漫画图像上的表现欠佳。 |

最佳实践为将`AniDet3.v3i`数据集与`v1.1`数据集结合用于模型训练。我们提供了[在线演示](https://huggingface.co/spaces/deepghs/anime_object_detection)。