README.md

---
license: mit
datasets:
- Zarxrax/Real-Anime-Segmentation
- Zarxrax/anime_image_segmentation
base_model:
- ZhengPeng7/BiRefNet_lite
---
This is a BiRefNet_lite anime background removal model. It is trained at 1024x1024 and is primarily for use on screenshots from actual anime videos.

![compare_1](https://cdn-uploads.huggingface.co/production/uploads/6338d2ff768611eccab78ed5/2H1RRRDapNlEuTlWaWaYO.png)
![compare_2](https://cdn-uploads.huggingface.co/production/uploads/6338d2ff768611eccab78ed5/hFJpcOHYPnJXWA8eloRHa.png)
![compare_3](https://cdn-uploads.huggingface.co/production/uploads/6338d2ff768611eccab78ed5/RMPduBl_S8R_Ae_nvJOeq.png)
![compare_4](https://cdn-uploads.huggingface.co/production/uploads/6338d2ff768611eccab78ed5/uxikdFJpWFbkPr90KZ4aD.png)
![compare_5](https://cdn-uploads.huggingface.co/production/uploads/6338d2ff768611eccab78ed5/J_i_Xor-pjK2AYC5jvdM3.png)