Search public pages, research tools, and SAVRN solutions.
Open-weight model · Image segmentation
by CAMPUS INSTITUTE DATA SCIENCE (CIDAS) CIDAS/clipseg-rd64-refined
CLIPSeg model with reduce dimension 64, refined (using a more complex convolution). It was introduced in the paper Image Segmentation Using Text and Image Prompts by Lüddecke et al. and first released in this repository.
What it takes to serve clipseg-rd64-refined (151M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.3 GB | 0.4 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.2 GB | 0.2 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
Give it a picture and a phrase, or a picture and a reference image, and it returns a segmentation mask. CIDAS built clipseg-rd64-refined for zero-shot and one-shot image segmentation from text and image prompts, per Lüddecke et al., arXiv:2112.10003. The text side takes at most 77 tokens, so prompts are short labels, not descriptions. The 151M parameters are stored in float32, which is why the download is 1.2 GB while the 16-bit run needs only 0.4 GB. The cheapest Index setup is one MI300X with 192 GB at $1.85 per hour.
Apache 2.0 covers commercial use, modification and redistribution, with notices kept. The rd64 in the name is a reduce dimension of 64, and refined means the variant with the more complex convolution, so match the checkpoint to what your pipeline was tuned against. Released November 1, 2022, last updated December 11, 2024, so confirm the revision.
By CAMPUS INSTITUTE DATA SCIENCE (CIDAS), published under apache-2.0, revision 999e0328d9e1.
CLIPSeg model with reduce dimension 64, refined (using a more complex convolution). It was introduced in the paper Image Segmentation Using Text and Image Prompts by Lüddecke et al. and first released in this repository. This model is intended for zero-shot and one-shot image segmentation. Refer to the documentation.
CLIPSeg model with reduce dimension 64, refined (using a more complex convolution). It was introduced in the paper Image Segmentation Using Text and Image Prompts by Lüddecke et al. and first released in this repository.
This model is intended for zero-shot and one-shot image segmentation.
Refer to the documentation.
10 files, 1.2 GB in total. The weights are 2 files totalling 1.2 GB in bin, safetensors.
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 603.0 MB | d00ca85d6b85 |
| pytorch_model.bin | Weights | 603.1 MB | dd9308225b83 |
| config.json | Configuration | 4.7 KB | — |
| preprocessor_config.json | Configuration | 380 B | — |
| special_tokens_map.json | Configuration | 472 B | — |
| README.md | Documentation | 596 B | — |
| .gitattributes | Repository | 1.4 KB | — |
| merges.txt | Tokenizer | 524.6 KB | — |
| tokenizer_config.json | Tokenizer | 974 B | — |
| vocab.json | Tokenizer | 1.1 MB | — |
Released by CAMPUS INSTITUTE DATA SCIENCE (CIDAS) through its official repository on Hugging Face. Read the license.
| Precision | Weights in memory |
|---|---|
| As published | 1.2 GB |
| 16-bit | 0.3 GB |
| 8-bit | 0.2 GB |
| 4-bit | 0.1 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
About 0.4 GB at 16-bit and 0.1 GB at 4-bit: the weights (151M parameters) plus a working margin. A long context needs more.
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Yes. clipseg-rd64-refined is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.
77 tokens, from the maximum position embeddings in its published configuration.
CLIPSeg model with reduce dimension 64. It was introduced in the paper Image Segmentation Using Text and Image Prompts by Lüddecke et al. and first released in this repository. This model is intended for zero-shot and one-shot image segmentation. Refer to the documentation.
Mask2Former model trained on ADE20k semantic segmentation (base-sized version, Swin backbone). It was introduced in the paper Masked-attention Mask Transformer for Universal Image Segmentation and first released in this repository. Disclaimer: The team releasing Mask2Former did not write a model card for this model so this model card has been written by the Hugging Face team. Mask2Former addresses instance, semantic and panoptic segmentation with the same paradigm: by predicting a set of masks and corresponding labels. Hence, all 3 tasks are treated as if they were instance segmentation. Mask2Former outperforms the previous SOTA, MaskFormer both in terms of performance an efficiency by (i)…
Mask2Former model trained on COCO panoptic segmentation (base-sized version, Swin backbone). It was introduced in the paper Masked-attention Mask Transformer for Universal Image Segmentation and first released in this repository. Disclaimer: The team releasing Mask2Former did not write a model card for this model so this model card has been written by the Hugging Face team. Mask2Former addresses instance, semantic and panoptic segmentation with the same paradigm: by predicting a set of masks and corresponding labels. Hence, all 3 tasks are treated as if they were instance segmentation. Mask2Former outperforms the previous SOTA, MaskFormer both in terms of performance an efficiency by (i)…
BEN2 (Background Erase Network) introduces a novel approach to foreground segmentation through its innovative Confidence Guided Matting (CGM) pipeline. The architecture employs a refiner network that targets and processes pixels where the base model exhibits lower confidence levels, resulting in more precise and reliable matting results. This model is built on BEN: BEN2 was trained on the DIS5k and our 22K proprietary segmentation dataset. Our enhanced model delivers superior performance in hair matting, 4K processing, object segmentation, and edge refinement. Our Base model is open source. To try the full model through our free web demo or integrate BEN2 into your project with our API…
Mask2Former model trained on Cityscapes semantic segmentation (large-sized version, Swin backbone). It was introduced in the paper Masked-attention Mask Transformer for Universal Image Segmentation and first released in this repository. Disclaimer: The team releasing Mask2Former did not write a model card for this model so this model card has been written by the Hugging Face team. Mask2Former addresses instance, semantic and panoptic segmentation with the same paradigm: by predicting a set of masks and corresponding labels. Hence, all 3 tasks are treated as if they were instance segmentation. Mask2Former outperforms the previous SOTA, MaskFormer both in terms of performance an efficiency by…
Mask2Former model trained on Mapillary Vistas semantic segmentation (large-sized version, Swin backbone). It was introduced in the paper Masked-attention Mask Transformer for Universal Image Segmentation and first released in this repository. Disclaimer: The team releasing Mask2Former did not write a model card for this model so this model card has been written by the Hugging Face team. Mask2Former addresses instance, semantic and panoptic segmentation with the same paradigm: by predicting a set of masks and corresponding labels. Hence, all 3 tasks are treated as if they were instance segmentation. Mask2Former outperforms the previous SOTA, MaskFormer both in terms of performance an…
