专为动漫角色优化背景移除模型，准确率提升至99.5%
ToonOut: Fine-tuned Background-Removal for Anime Characters
- 基于1228张动漫图像微调BiRefNet模型
- 动漫图像背景移除准确率从95.3%提升至99.5%
- 适合动漫创作、图像编辑等场景使用
当前先进的背景移除模型在真实图像上表现良好，但在动漫风格内容中因发丝、透明度等复杂特征而性能下降。为此，我们收集并标注了1,228张高质量动漫角色与物体图像，基于该数据集对开源的BiRefNet模型进行微调。结果表明，新提出的像素准确率（Pixel Accuracy）指标下，动漫图像背景移除准确率从95.3%提升至99.5%。代码、微调模型权重及数据集已公开：https://github.com/MatteoKartoon/BiRefNet。
原文摘要 · Abstract (English)
While state-of-the-art background removal models excel at realistic imagery, they frequently underperform in specialized domains such as anime-style content, where complex features like hair and transparency present unique challenges. To address this limitation, we collected and annotated a custom dataset of 1,228 high-quality anime images of characters and objects, and fine-tuned the open-sourced BiRefNet model on this dataset. This resulted in marked improvements in background removal accuracy for anime-style images, increasing from 95.3% to 99.5% for our newly introduced Pixel Accuracy metric. We are open-sourcing the code, the fine-tuned model weights, as well as the dataset at: https://github.com/MatteoKartoon/BiRefNet.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。