Video
Video Foreground Extractor
Use it ↗BiRefNet v2 provides high-resolution foreground extraction for complex objects like hair and transparent edges.
BiRefNet v2 is specialized for "dichotomous" segmentation, meaning it identifies and separates objects from complex backgrounds with high precision. This makes it particularly effective for e-commerce, green-screen replacement, and 3D compositing. This is achieved through a bilateral reference system that captures both global context and fine local details, enabling pixel-perfect knowledge of the visual environment.