Feyn releases FeyNoBg, an open-source AI background removal model and training library
Feyn, an AI company that helps businesses build custom models, has launched FeyNoBg, an automatic background removal model, along with NoBg, an open-source Python library for training and running it. The model was trained on over 26,000 images drawn from 10 datasets, with a focus on balancing two core skills: foreground identification and precise boundary tracing. To improve performance, the team expanded a key stage of the underlying BiRefNet architecture from 18 to 24 blocks after studying how each stage contributes to the task. FeyNoBg achieves top published scores on four out of eight benchmarks and comes within 2% of the leading model on the remaining four. NoBg was also built to address fragmented tooling in the image matting space, unifying preprocessing, training, and evaluation under a single Python interface.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in