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…
Open-weight model · Image segmentation
dinov3-pointing-pilot-v1
by Fire Viewer fireviewer/dinov3-pointing-pilot-v1
dinov3-pointing-pilot-v1 is an open-weight model for image segmentation from Fire Viewer, released under other. It has 94M parameters. At 16-bit it needs about 0.2 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 44 downloads a month.
Research archive, currently publicly visible on the Hub, of a five-epoch DINOv3 ViT-B/16 multi-task pilot for visible fire/smoke presence, segmentation, point localization and explicit visual abstention.
Parameters94M
Context—
Weights1.9 GB
Licenseother
AccessOpen weights
Monthly Downloads44
Runs On
What it takes to serve dinov3-pointing-pilot-v1 (94M 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 | 0.2 GB | 0.2 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.0 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 Oct 7, 2026.
dinov3-pointing-pilot-v1 on every accelerator the SAVRN Index prices, at every precision
Model Card
Research archive, currently publicly visible on the Hub, of a five-epoch DINOv3 ViT-B/16 multi-task pilot for visible fire/smoke presence, segmentation, point localization and explicit visual abstention. This repository preserves a completed, reloadable experiment. It is not a production model and not independently benchmarked. The retained release records describe a private pilot policy; current public visibility does not resolve that restriction or establish redistribution permission. The word “pilot” is part of the model identity and should not be removed in downstream documentation. The metrics below come from the campaign validation split. They are not an independent benchmark and must…
Excerpt from the card by Fire Viewer, licensed other.
Configuration
- Architecture
- DinoV3MultiTaskModel
Identity and Version
- Repository
- fireviewer/dinov3-pointing-pilot-v1
- Publisher
- Fire Viewer
- Task
- Image segmentation
- Modality
- Image
- Library
- pytorch
- Parameters
- 94M parameters
- Languages
- Not stated by the source
- Revision
- 0de594fef03cbd25d098c654dc3a177682ea8705
- First published
- 2026-09-08
- Last updated
- 2026-10-05
Files and Weights
20 files, 1.9 GB in total. The weights are 3 files totalling 1.9 GB in pt, safetensors.
Weights3 files · 1.9 GB
Configuration11 files · 118.0 KB
Documentation2 files · 7.9 KB
Other3 files · 143.2 KB
Repository1 file · 1.5 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| checkpoints/best.pt | Weights | 376.6 MB | 00a880ea4841 |
| checkpoints/last.pt | Weights | 1.1 GB | 77bff6d09d91 |
| model.safetensors | Weights | 376.6 MB | dc82b65fb681 |
| artifact-manifest.json | Configuration | 1.8 KB | — |
| config.json | Configuration | 498 B | — |
| dinov3_adapter.py | Configuration | 71.7 KB | — |
| reports/accepted-smoke-receipt.json | Configuration | 10.1 KB | — |
| reports/preflight-report.json | Configuration | 5.7 KB | — |
| reports/sampling-report.json | Configuration | 1.4 KB | — |
| reports/training-plan.json | Configuration | 5.3 KB | — |
| reports/training-result.json | Configuration | 19.7 KB | — |
| reports/wandb-sync-receipt.json | Configuration | 82 B | — |
| safetensors-conversion.json | Configuration | 625 B | — |
| validation-summary.json | Configuration | 1.1 KB | — |
| README.md | Documentation | 7.0 KB | — |
| RIGHTS_AND_ATTRIBUTION.md | Documentation | 881 B | — |
| assets/training-curves.png | Other | 83.4 KB | — |
| reports/metrics.csv | Other | 59.8 KB | — |
| requirements.txt | Other | 56 B | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- other
- Access
- Open weights, no gate
- Download size
- 1.9 GB
Download from Fire Viewer
Released by Fire Viewer through its official repository on Hugging Face. Read the license.
Built From
- Derived from facebook/dinov3-vitb16-pretrain-lvd1689m
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 1.9 GB |
| 16-bit | 0.2 GB |
| 8-bit | 0.1 GB |
| 4-bit | 0.0 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About dinov3-pointing-pilot-v1
How much GPU memory does dinov3-pointing-pilot-v1 need?
About 0.2 GB at 16-bit and 0.1 GB at 4-bit: the weights (94M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run dinov3-pointing-pilot-v1 on?
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.
What license is dinov3-pointing-pilot-v1 released under?
other, as its publisher declares it. Read the license text before commercial use.
Similar Models
Semantic segmentation model fine-tuned from nvidia/mit-b5 with CelebAMask-HQ for face parsing. For additional options, see the Transformers Segformer docs. Exhaustive list of labels can be extracted from config.json. Since p5.js uses an animation loop abstraction, we need to take care loading the model and making predictions. While the capabilities of computer vision models are impressive, they can also reinforce or exacerbate social biases. The CelebAMask-HQ dataset used for fine-tuning is large but not necessarily perfectly diverse or representative. Also, they are images of.... just celebrities.
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)…
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 ADE20k semantic segmentation (small-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 instance segmentation (small-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)…