10 models · ranked by HuggingFace downloads
clipseg-rd64-refined targets image segmentation and is shipped as an open-weight, self-hostable checkpoint. Permissive Apache 2.0 terms let clipseg-rd64-refined go straight into commercial pipelines. Treat clipseg-rd64-refined's published metrics as a starting point and validate against your workload.
BiRefNet is an openly licensed image segmentation model. BiRefNet is MIT-licensed, clearing it for closed-source and paid products. Evaluate BiRefNet on your own data before trusting it in production.
RMBG-2.0 targets image segmentation and is shipped as an open-weight, self-hostable checkpoint. Licensing for RMBG-2.0 is unspecified or custom — clear it before commercial use. RMBG-2.0 is community-maintained, so track upstream changes and pin a known-good revision.
As an open-weight model, RMBG-1.4 focuses on image segmentation. RMBG-1.4 lists a non-standard license, so confirm permissions before deployment. RMBG-1.4 ships without a hosted SLA, so budget for self-managed deployment and monitoring.
segformer_b2_clothes is an open-weight image segmentation model. Licensing for segformer_b2_clothes is unspecified or custom — clear it before commercial use. Evaluate segformer_b2_clothes on your own data before trusting it in production.
mask2former-swin-large-ade-semantic is Meta's Mask2Former architecture with a Swin-Large backbone, fine-tuned for semantic segmentation on the ADE20K dataset. It unifies panoptic, instance, and semantic segmentation under a single masked-attention transformer framework (arxiv:2112.01527).
SegFormer-B2 fine-tuned on the ADE20K scene parsing dataset at 512×512 resolution. SegFormer is a transformer-based semantic segmentation model that combines hierarchical transformer encoders with a lightweight MLP decoder. This checkpoint is optimized for the 150-class ADE20K scene understanding benchmark.
SegFormer-B0 fine-tuned on ADE20K at 512×512 resolution, providing semantic segmentation of 150 scene categories. The B0 variant is the smallest in the SegFormer family, optimized for speed over maximum accuracy.
A SegFormer-based face parsing model from jonathandinu that segments facial regions (hair, eyes, nose, mouth, skin, etc.) from portrait images. Trained on CelebAMask-HQ, it outputs per-pixel class labels for 19 semantic facial regions.
segformer-b0-finetuned-ade-512-512 is an open-weight checkpoint for image segmentation, distributed on the HuggingFace Hub. Evaluate segformer-b0-finetuned-ade-512-512 on your own data before trusting it in production.