ToonOut: Fine-tuned Background-Removal for Anime Characters
This work addresses a domain-specific problem for anime content creators and enthusiasts, representing an incremental improvement by applying an existing method to new data.
The paper tackled the problem of background removal for anime-style images, where existing models underperform due to complex features like hair and transparency, by fine-tuning the BiRefNet model on a custom dataset of 1,228 anime images, resulting in an increase in Pixel Accuracy from 95.3% to 99.5%.
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.