libreyolo
LibreYOLO is a MIT licensed open source computer vision library
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Check it deeper on your machine
It has changed since its scan on 11 Nov 2026.
As a Claude Code plugin, install it and then ask Claude Code:
/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/LibreYOLO/libreyolo with scanskillsafety
Or from a terminal, clone it and run it:
git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/LibreYOLO/libreyolo'
It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.
Badge for your README
Markdown
[](/listings/libreyolo--libreyolo)HTML
<a href="/listings/libreyolo--libreyolo"><img src="/api/listings/libreyolo--libreyolo/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
pip install libreyolo
from libreyolo import LibreYOLO, SAMPLE_IMAGE
model = LibreYOLO("LibreYOLO9t.pt")
result = model(SAMPLE_IMAGE, save=True)<details> <summary><b>Optional extras</b></summary>
<br>
The base install covers YOLOv9 and the other core detectors, training, and inference. Add an extra when you need a heavier family or an export backend. Comma-separate to combine, for example `pip install "libreyolo[rfdetr,hub-kernels]"`.
| Group | Extras | | --- | --- | | Export | `onnx`, `tensorrt`, `openvino`, `coreml`, `coreai`, `tflite` (alias `litert`), `ncnn`, `mnn`, `paddle`, `executorch` | | Serving | `triton` | | Models | `rfdetr`, `vlm`, `sam`, `openvocab`, `clip`, `siglip2`, `eomt`, `midas`, `modus`, `sensenova`, `gaze` | | Training | `lora`, `plots`, `tensorboard`, `mlflow`, `wandb`, `comet`, `clearml`, `neptune`, `dvclive` | | Speed | `fast-eval`, `hub-kernels` | | Sources | `stream` | | Everything | `pip install "libreyolo[all]"` |
`executorch`, `coreai` and `neptune` are deliberately left out of `all`: they pin torch or protobuf in ways that would drag the rest of the environment with them. Full list and per-backend notes in the [install guide](https://www.libreyolo.com/docs/install).
</details>
<details> <summary><b>Install from source</b></summary>
<br>
git clone https://github.com/LibreYOLO/libreyolo.git cd libreyolo pip install -e .
A plain clone checks out `release`, the stable branch matching the published package. For unreleased work, `git checkout dev`.
</details>
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 313 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.
- <div align="center">
- <img src="docs/readme/banner.png" alt="LibreYOLO, MIT licensed computer vision. pip install libreyolo" width="100%">
- <br>
- LibreYOLO is an MIT licensed computer vision library to train, run, and export computer vision models.
- [](https://pypi.org/project/libreyolo/)
- [](https://pepy.tech/projects/libreyolo)
- [](LICENSE)
- [](https://www.libreyolo.com/docs)
- [](https://huggingface.co/LibreYOLO)
- [](https://colab.research.google.com/github/LibreYOLO/libreyolo/blob/release/notebooks/quickstart.ipynb)
- [](https://www.greptile.com/?utm_source=oss_badge&utm_medium=readme&utm_campaign=greptile_for_open_source)
- [English](README.md) | [简体中文](README.zh-CN.md)
- </div>
- > ⭐ **Support LibreYOLO.** The best way to help is to **star the repo**. Feel free to [open an issue](https://github.com/LibreYOLO/libreyolo/issues/new) if you encounter problems or have suggestions, and code contributions are very welcome (see [CONTRIBUTING.md](CONTRIBUTING.md)). If you want to contribute economically, that is very welcome: see the [sponsorship program](SPONSORS.md).
- <p align="center"><img src="libreyolo/assets/guggenheim-bilbao_result.jpg" width="520" alt="LibreYOLO detection on the Guggenheim Bilbao image"></p>
- ## Install
- ```bash
- pip install libreyolo
- ```
- <details>
- <summary><b>Optional extras</b></summary>
- <br>
- The base install covers YOLOv9 and the other core detectors, training, and
- inference. Add an extra when you need a heavier family or an export backend.
- Comma-separate to combine, for example `pip install "libreyolo[rfdetr,hub-kernels]"`.
- | Group | Extras |
- | --- | --- |
- | Export | `onnx`, `tensorrt`, `openvino`, `coreml`, `coreai`, `tflite` (alias `litert`), `ncnn`, `mnn`, `paddle`, `executorch` |
- | Serving | `triton` |
- | Models | `rfdetr`, `vlm`, `sam`, `openvocab`, `clip`, `siglip2`, `eomt`, `midas`, `modus`, `sensenova`, `gaze`, `gen` |
- | Training | `lora`, `plots`, `tensorboard`, `mlflow`, `wandb`, `comet`, `clearml`, `neptune`, `dvclive` |
- | Speed | `fast-eval`, `hub-kernels` |
- | Sources | `stream` |
- | Everything | `pip install "libreyolo[all]"` |
- `executorch`, `coreai`, `paddle` and `neptune` are deliberately left out of
- `all`: they pin torch, onnx or protobuf in ways that would drag the rest of the
- environment with them. `fast-eval` and `hub-kernels` are not in `all` either.
- Full list and per-backend notes in the
- [install guide](https://www.libreyolo.com/docs/install).
- </details>
- <details>
- <summary><b>Install from source</b></summary>
- <br>
- ```bash
- git clone https://github.com/LibreYOLO/libreyolo.git
- cd libreyolo
- pip install -e .
- ```
- A plain clone checks out `release`, the stable branch matching the published
- package. For unreleased work, `git checkout dev`.
- </details>
- <details>
- <summary><b>If huggingface.co is blocked (mainland China)</b></summary>
- <br>
- Weights download from Hugging Face. LibreYOLO honours the standard
- `HF_ENDPOINT` variable, so you can point it at a mirror such as
- [hf-mirror.com](https://hf-mirror.com), a community mirror not run by
- Hugging Face or LibreYOLO:
- ```bash
- export HF_ENDPOINT=https://hf-mirror.com
- ```
- Or download the file manually, for example
- `https://hf-mirror.com/LibreYOLO/LibreYOLO9t/resolve/main/LibreYOLO9t.pt`,
- and pass its local path: `LibreYOLO("path/to/LibreYOLO9t.pt")`.
- </details>
- ## Train
- ```python
- from libreyolo import LibreYOLO
- model = LibreYOLO("LibreYOLO9t.pt")
- model.train(data="dataset.yaml", epochs=100, imgsz=640)
- ```
- ```bash
- libreyolo train --model yolo9-t --data dataset.yaml --epochs 100
- ```
- See the [training guide](https://www.libreyolo.com/docs/train) to read about
- multi-GPU, LoRA, layer freezing, distillation, from-scratch training,
- resume, mixed precision, early stopping and logging.
- ## Inference
- ```python
- from libreyolo import LibreYOLO, SAMPLE_IMAGE
- model = LibreYOLO("LibreYOLO9t.pt")
- result = model(SAMPLE_IMAGE, save=True)
- ```
- Load any model straight from Hugging Face by its repo name: `LibreYOLO("LibreYOLO/LibreRFDETRn")`.
- Sources are not just files. Point it at a webcam, an RTSP stream, a video, a
- directory, a YouTube URL or your screen:
- ```bash
- libreyolo predict --model yolo9-t --source 0 --show # webcam
- libreyolo predict --model yolo9-t --source rtsp://camera/1 # network camera
- libreyolo predict --model yolo9-t --source screen # screen capture
- ```
- See the [inference guide](https://www.libreyolo.com/docs/predict) to read about
- sources, results, thresholds and performance.
- ## Export
- Twelve formats: ONNX, TorchScript, TensorRT, OpenVINO, CoreML, Core AI, TFLite
- (LiteRT), NCNN, MNN, RKNN, Paddle and ExecuTorch. Plus NVIDIA Triton serving
- and DeepStream config generation.
- ```bash
- libreyolo export --model yolo9-t --format onnx
- ```
- See the [export guide](https://www.libreyolo.com/docs/export) for each
- format. Support varies by family and task, see the
- [export matrix](https://www.libreyolo.com/docs/reference/export-matrix).
- ## Documentation
- - [Docs](https://www.libreyolo.com/docs) covers install, tasks, models, training, prediction, export and the CLI
- - [Benchmarks](https://www.visionanalysis.org/) on Vision Analysis
- - [CHANGELOG.md](CHANGELOG.md) for what changed
- ## Complete list of models
- LibreYOLO is organized by task, such as detection, segmentation or
- classification. Each task has several models, and each model comes in several
- variants. For example, `LibreYOLO9t` is task detection, model YOLO9 and
- variant `t`, which means tiny.
- ```mermaid
- %%{init: {"flowchart": {"htmlLabels": true, "wrappingWidth": 270, "nodeSpacing": 14, "rankSpacing": 10, "padding": 10, "diagramPadding": 6, "curve": "basis"}}}%%
- flowchart TB
- ROOT(["<big><big><b>LibreYOLO</b></big></big><br/><small>151 model families, 23 tasks</small>"])
- subgraph C0[" "]
- direction TB
- H0(["<b>SEE</b> (76)"])
- T0_0_0["<big><b>YOLO-style detectors</b></big> <small>(12)</small><br/><b>YOLOv9</b>, <b>YOLOX</b>, <b>YOLO-#8288;NAS</b>, YOLOv7, YOLOv4, YOLOv3, YOLOv2, YOLOv1, RTMDet, PicoDet, PP-#8288;YOLOE, EfficientDet"]:::c0
- T0_0_1["<big><b>DETR-style detectors</b></big> <small>(13)</small><br/><b>RF-#8288;DETR</b>, <b>D-#8288;FINE</b>, <b>DEIM</b>, DEIMv2, RT-#8288;DETR, RT-#8288;DETRv2, RT-#8288;DETRv4, LW-#8288;DETR, DETR, Deformable#nbsp;DETR, DINO-#8288;DETR, EdgeCrafter, GTR"]:::c0
- T0_0_2["<big><b>Classic CNN detectors</b></big> <small>(6)</small><br/><b>Faster#nbsp;R-#8288;CNN</b>, Mask#nbsp;R-#8288;CNN, RetinaNet, SSD, FCOS, CenterNet"]:::c0
- T0_0_3["<big><b>Open-vocabulary detectors</b></big> <small>(6)</small><br/><b>Grounding#nbsp;DINO</b>, <b>OWLv2</b>, OMDet-#8288;Turbo, OV-#8288;DEIM, LocateAnything, LibreMODUS"]:::c0
- T0_0_4["<big><b>VLM detectors</b></big> <small>(12)</small><br/><b>Qwen3-#8288;VL</b>, <b>Florence-#8288;2</b>, Gemma#nbsp;4, Moondream, InternVL3, Kosmos-#8288;2, LFM2-#8288;VL, North#nbsp;Micro#nbsp;Vision, SmolVLM2, WeVisOne-#8288;VL, SenseNova-#8288;Vision, Falcon#nbsp;Perception"]:::c0
- T0_0_5["<big><b>Specialised detectors</b></big> <small>(3)</small><br/><b>TinyFormer</b>, Dome-#8288;DETR, PP-#8288;DocLayoutV3"]:::c0
- H0 ~~~ T0_0_0 ~~~ T0_0_1 ~~~ T0_0_2 ~~~ T0_0_3 ~~~ T0_0_4 ~~~ T0_0_5
- T0_1_0["<big><b>Instance segmentation</b></big> <small>(17)</small><br/><b>RF-#8288;DETR</b>, <b>SAM#nbsp;3</b>, <b>SAM#nbsp;2</b>, <b>EoMT</b>, D-#8288;FINE, EdgeCrafter, EdgeTAM, Falcon#nbsp;Perception, GTR, Mask#nbsp;R-#8288;CNN, MobileSAM, PicoSAM3, RTMDet, SAM, SAM#nbsp;3.1, SenseNova-#8288;Vision, WeVisOne-#8288;VL"]:::c0
- T0_1_1["<big><b>Semantic segmentation</b></big> <small>(10)</small><br/><b>SegFormer</b>, <b>EoMT</b>, <b>DeepLabv3</b>, DINOv2, FCN, GTR, LingBot-#8288;Vision, PIDNet, PP-#8288;LiteSeg, U-#8288;Net"]:::c0
- T0_1_2["<big><b>Panoptic segmentation</b></big> <small>(2)</small><br/>EoMT, SenseNova-#8288;Vision"]:::c0
- T0_1_3["<big><b>Oriented boxes</b></big> <small>(5)</small><br/><b>RF-#8288;DETR</b>, <b>RiO-#8288;DETRv2</b>, GTR, RT-#8288;DETRv2, YOLO-#8288;NAS"]:::c0
- T0_1_4["<big><b>Pose</b></big> <small>(7)</small><br/><b>RF-#8288;DETR</b>, <b>HRNet</b>, <b>YOLO-#8288;NAS</b>, DEKR, EdgeCrafter, GTR, SenseNova-#8288;Vision"]:::c0
- T0_1_5["<big><b>Pointing</b></big> <small>(9)</small><br/><b>Molmo2</b>, <b>Qwen3-#8288;VL#nbsp;grounding</b>, <b>Moondream</b>, Florence-#8288;2#nbsp;grounding, FOMO, LocateAnything, SenseNova-#8288;Vision, ShowUI, WeVisOne-#8288;VL"]:::c0
- H0 ~~~ T0_1_0 ~~~ T0_1_1 ~~~ T0_1_2 ~~~ T0_1_3 ~~~ T0_1_4 ~~~ T0_1_5
- end
- subgraph C1[" "]
- direction TB
- H1(["<b>UNDERSTAND</b> (26)"])
- T1_0_0["<big><b>Classification</b></big> <small>(15)</small><br/><b>DINOv2</b>, <b>ConvNeXt#nbsp;V2</b>, <b>SigLIP2</b>, <b>ViT</b>, AlexNet, CLIP, ConvNeXt, DeiT, EfficientNetV2, MobileNetV4, Perception#nbsp;Encoder, ResNet, Swin#nbsp;Transformer, V-#8288;JEPA#nbsp;2, VGG"]:::c1
- T1_0_1["<big><b>Embeddings</b></big> <small>(8)</small><br/><b>DINOv2</b>, <b>SigLIP2</b>, <b>CLIP</b>, LeVJEPA, LibreFaceEmbedder, Perception#nbsp;Encoder, Qwen3-#8288;VL-#8288;Embedding, V-#8288;JEPA#nbsp;2"]:::c1
- T1_0_2["<big><b>OCR</b></big> <small>(7)</small><br/><b>PP-#8288;OCRv5</b>, <b>GLM-#8288;OCR</b>, <b>DeepSeek-#8288;OCR</b>, Falcon#nbsp;OCR, PaddleOCR-#8288;VL, SenseNova-#8288;Vision, WeVisOne-#8288;VL"]:::c1
- T1_0_3["<big><b>Gaze</b></big> <small>(1)</small><br/>L2CS-#8288;Net"]:::c1
- H1 ~~~ T1_0_0 ~~~ T1_0_1 ~~~ T1_0_2 ~~~ T1_0_3
- end
- subgraph C2[" "]
- direction TB
- H2(["<b>GEOMETRY</b> (28)"])
- T2_0_0["<big><b>Depth</b></big> <small>(11)</small><br/><b>Depth#nbsp;Anything#nbsp;3</b>, <b>MoGe-#8288;3</b>, <b>Video#nbsp;Depth#nbsp;Anything</b>, Depth#nbsp;Anything#nbsp;V2, GTR, LibreMODUS, Marigold#nbsp;V2, MiDaS, SenseNova-#8288;Vision, Speridlabs#nbsp;Iris-#8288;Z, ZipDepth"]:::c2
- T2_0_1["<big><b>Surface normals</b></big> <small>(4)</small><br/>LibreMODUS, Marigold#nbsp;V2, MoGe-#8288;2, MoGe-#8288;3"]:::c2
- T2_0_2["<big><b>Albedo</b></big> <small>(1)</small><br/>Marigold#nbsp;V2"]:::c2
- T2_0_3["<big><b>Edges</b></big> <small>(3)</small><br/>DexiNed, LibreMODUS, TEED"]:::c2
- H2 ~~~ T2_0_0 ~~~ T2_0_1 ~~~ T2_0_2 ~~~ T2_0_3
- T2_1_0["<big><b>Optical flow</b></big> <small>(8)</small><br/><b>SEA-#8288;RAFT</b>, <b>WAFT</b>, <b>RAFT</b>, DPFlow, FlowSeek, MEMFOF, NeuFlow#nbsp;v2, UniMatch"]:::c2
- T2_1_1["<big><b>3D detection</b></big> <small>(4)</small><br/>3D-#8288;MOOD, DetAny3D, FCOS3D, WildDet3D"]:::c2
- T2_1_2["<big><b>Body mesh</b></big> <small>(2)</small><br/>InstantHMR, SAM#nbsp;3D#nbsp;Body"]:::c2
- H2 ~~~ T2_1_0 ~~~ T2_1_1 ~~~ T2_1_2
- end
- subgraph C3[" "]
- direction TB
- H3(["<b>RESTORE AND CREATE</b> (26)"])
- T3_0_0["<big><b>Restoration</b></big> <small>(7)</small><br/><b>Real-#8288;ESRGAN</b>, <b>SwinIR</b>, <b>LaMa</b>, DDColor, HVI-#8288;CIDNet, NAFNet, QuickSRNet"]:::c3
- T3_0_1["<big><b>Background removal</b></big> <small>(5)</small><br/><b>BiRefNet</b>, <b>BEN2</b>, <b>ViTMatte</b>, FeyNobg, WeVisOne-#8288;VL"]:::c3
- T3_0_2["<big><b>Image generation</b></big> <small>(6)</small><br/><b>FLUX.2#nbsp;klein</b>, <b>Speridlabs#nbsp;Iris-#8288;Z</b>, <b>SANA-#8288;Sprint</b>, Open#nbsp;Vision#nbsp;Banana, PRX, PRXPixel"]:::c3
- T3_0_3["<big><b>3D generation</b></big> <small>(8)</small><br/><b>TRELLIS.2</b>, <b>Speridlabs#nbsp;Idol</b>, <b>TripoSR</b>, CRM, Direct3D-#8288;S2, LGM, TRELLIS, TripoSplat"]:::c3
- H3 ~~~ T3_0_0 ~~~ T3_0_1 ~~~ T3_0_2 ~~~ T3_0_3
- end
- subgraph C4[" "]
- direction TB
- H4(["<b>ROBOTICS</b> (3)"])
- T4_0_0["<big><b>Robot policies</b></big> <small>(3)</small><br/>ACT, Diffusion#nbsp;Policy, SmolVLA"]:::c4
- H4 ~~~ T4_0_0
- end
- ROOT --> H0
- ROOT --> H1
- ROOT --> H2
- ROOT --> H3
- ROOT --> H4
- style ROOT fill:#1f2328,stroke:#8b949e,stroke-width:1px,color:#ffffff,font-size:20px
- classDef c0 fill:#fff8c5,stroke:#bf8700,stroke-width:1px,color:#1f2328,font-size:15px,text-align:left,rx:10,ry:10
- style C0 fill:transparent,stroke:#bf8700,stroke-width:2px,rx:14,ry:14
- style H0 fill:#bf8700,stroke:#bf8700,color:#ffffff,font-size:17px
- classDef c1 fill:#dafbe1,stroke:#1a7f37,stroke-width:1px,color:#1f2328,font-size:15px,text-align:left,rx:10,ry:10
- style C1 fill:transparent,stroke:#1a7f37,stroke-width:2px,rx:14,ry:14
- style H1 fill:#1a7f37,stroke:#1a7f37,color:#ffffff,font-size:17px
- classDef c2 fill:#fbefff,stroke:#8250df,stroke-width:1px,color:#1f2328,font-size:15px,text-align:left,rx:10,ry:10
- style C2 fill:transparent,stroke:#8250df,stroke-width:2px,rx:14,ry:14
- style H2 fill:#8250df,stroke:#8250df,color:#ffffff,font-size:17px
- classDef c3 fill:#ffeff7,stroke:#bf3989,stroke-width:1px,color:#1f2328,font-size:15px,text-align:left,rx:10,ry:10
- style C3 fill:transparent,stroke:#bf3989,stroke-width:2px,rx:14,ry:14
- style H3 fill:#bf3989,stroke:#bf3989,color:#ffffff,font-size:17px
- classDef c4 fill:#ddf4ff,stroke:#0969da,stroke-width:1px,color:#1f2328,font-size:15px,text-align:left,rx:10,ry:10
- style C4 fill:transparent,stroke:#0969da,stroke-width:2px,rx:14,ry:14
- style H4 fill:#0969da,stroke:#0969da,color:#ffffff,font-size:17px
- ```
- <details open>
- <summary><b>All tasks and models</b></summary>
- <br>
- | Task | Models |
- | --- | --- |
- | **Detection** | [YOLOv9](https://www.libreyolo.com/docs/models/yolov9), [RF-DETR](https://www.libreyolo.com/docs/models/rf-detr), [YOLOX](https://www.libreyolo.com/docs/models/yolox), [YOLO-NAS](https://www.libreyolo.com/docs/models/yolo-nas), [D-FINE](https://www.libreyolo.com/docs/models/d-fine), [DEIM](https://www.libreyolo.com/docs/models/deim), [RT-DETR v1/v2/v4](https://www.libreyolo.com/docs/models/rt-detr), [RTMDet](https://www.libreyolo.com/docs/models/rtmdet), [PicoDet](https://www.libreyolo.com/docs/models/picodet), [PP-YOLOE](https://www.libreyolo.com/docs/models/ppyoloe), [YOLOv7](https://www.libreyolo.com/docs/models/yolov7), [EfficientDet](https://www.libreyolo.com/docs/models/efficientdet), and the classics: [DETR](https://www.libreyolo.com/docs/models/detr), [Deformable DETR](https://www.libreyolo.com/docs/models/deformable-detr), [DINO-DETR](https://www.libreyolo.com/docs/models/dino-detr), [LW-DETR](https://www.libreyolo.com/docs/models/lw-detr), [Faster R-CNN](https://www.libreyolo.com/docs/models/faster-rcnn), [RetinaNet](https://www.libreyolo.com/docs/models/retinanet), [SSD](https://www.libreyolo.com/docs/models/ssd), [FCOS](https://www.libreyolo.com/docs/models/fcos), [CenterNet](https://www.libreyolo.com/docs/models/centernet) |
- | **Tiny objects** | [TinyFormer](https://www.libreyolo.com/docs/models/tinyformer), [Dome-DETR](https://www.libreyolo.com/docs/models/dome-detr) (aerial, drone, remote sensing) |
- | **Instance segmentation** | [RF-DETR](https://www.libreyolo.com/docs/models/rf-detr), [RTMDet](https://www.libreyolo.com/docs/models/rtmdet), [D-FINE](https://www.libreyolo.com/docs/models/d-fine), [Mask R-CNN](https://www.libreyolo.com/docs/models/mask-rcnn) |
- | **Promptable segmentation** | [SAM](https://www.libreyolo.com/docs/models/sam), [SAM 2](https://www.libreyolo.com/docs/models/sam-2), [SAM 3](https://www.libreyolo.com/docs/models/sam-3), [MobileSAM](https://www.libreyolo.com/docs/models/mobilesam), [EdgeTAM](https://www.libreyolo.com/docs/models/edgetam), [PicoSAM3](https://www.libreyolo.com/docs/models/picosam3) |
- | **Semantic segmentation** | [SegFormer](https://www.libreyolo.com/docs/models/segformer), [PIDNet](https://www.libreyolo.com/docs/models/pidnet), [PP-LiteSeg](https://www.libreyolo.com/docs/models/ppliteseg), [U-Net](https://www.libreyolo.com/docs/models/unet), [DeepLabv3](https://www.libreyolo.com/docs/models/deeplabv3), [FCN](https://www.libreyolo.com/docs/models/fcn), [LingBot-Vision](https://www.libreyolo.com/docs/models/lingbot-vision), [DINOv2](https://www.libreyolo.com/docs/models/dinov2), [EoMT](https://www.libreyolo.com/docs/models/eomt) |
- | **Panoptic segmentation** | [EoMT](https://www.libreyolo.com/docs/models/eomt) |
- | **Pose** | [RF-DETR](https://www.libreyolo.com/docs/models/rf-detr), [YOLO-NAS](https://www.libreyolo.com/docs/models/yolo-nas), [HRNet](https://www.libreyolo.com/docs/models/hrnet), [DEKR](https://www.libreyolo.com/docs/models/dekr), [EC](https://www.libreyolo.com/docs/models/edgecrafter) |
- | **Oriented boxes** | [RF-DETR](https://www.libreyolo.com/docs/models/rf-detr), [RT-DETRv2](https://www.libreyolo.com/docs/models/rt-detr), RiO-DETRv2, [YOLO-NAS-R](https://www.libreyolo.com/docs/models/yolo-nas) |
- | **3D detection** | [WildDet3D](https://www.libreyolo.com/docs/models/wilddet3d), [3D-MOOD](https://www.libreyolo.com/docs/models/3d-mood), [FCOS3D](https://www.libreyolo.com/docs/models/fcos3d), [DetAny3D](https://www.libreyolo.com/docs/models/detany3d) (separate runtime) |
- | **Classification** | [MobileNetV4](https://www.libreyolo.com/docs/models/mobilenetv4), [ConvNeXt](https://www.libreyolo.com/docs/models/convnext), [EfficientNetV2](https://www.libreyolo.com/docs/models/efficientnetv2), [ResNet](https://www.libreyolo.com/docs/models/resnet), [ViT](https://www.libreyolo.com/docs/models/vit), [Swin](https://www.libreyolo.com/docs/models/swin), [DeiT](https://www.libreyolo.com/docs/models/deit), [VGG](https://www.libreyolo.com/docs/models/vgg), [AlexNet](https://www.libreyolo.com/docs/models/alexnet), [CLIP](https://www.libreyolo.com/docs/models/clip), [SigLIP2](https://www.libreyolo.com/docs/models/siglip2), [DINOv2](https://www.libreyolo.com/docs/models/dinov2) |
- | **Depth** | [Depth Anything 3](https://www.libreyolo.com/docs/models/depth-anything-3), [Depth Anything V2](https://www.libreyolo.com/docs/models/depth-anything-v2), Video Depth Anything, [ZipDepth](https://www.libreyolo.com/docs/models/zipdepth), [MiDaS](https://www.libreyolo.com/docs/models/midas), MoGe-3, Speridlabs Iris-Z |
- | **Surface normals** | MoGe-3, [MoGe-2](https://www.libreyolo.com/docs/models/moge-2) |
- | **Depth, normals and albedo** | [Marigold V2](https://www.libreyolo.com/docs/models/marigold-v2) |
- | **Edges** | [DexiNed](https://www.libreyolo.com/docs/models/dexined), [TEED](https://www.libreyolo.com/docs/models/teed) |
- | **Optical flow** | SEA-RAFT, NeuFlow v2, RAFT, WAFT, FlowSeek, DPFlow, UniMatch, MEMFOF |
- | **Embeddings** | [LibreFaceEmbedder](https://www.libreyolo.com/docs/models/librefacerec), [CLIP](https://www.libreyolo.com/docs/models/clip), [SigLIP2](https://www.libreyolo.com/docs/models/siglip2), [Perception Encoder](https://www.libreyolo.com/docs/models/pe) (image, text, whole-video; also zero-shot classify), [DINOv2](https://www.libreyolo.com/docs/models/dinov2) |
- | **Video embeddings** | [V-JEPA 2](https://www.libreyolo.com/docs/models/vjepa2) (clip embedding and trainable classification probe), [LeVJEPA](https://www.libreyolo.com/docs/models/levjepa) (clip and patch embeddings; CC BY-NC 4.0 weights) |
- | **Body mesh** | [SAM 3D Body](https://www.libreyolo.com/docs/models/sam-3d-body), InstantHMR |
- | **Restoration** | [DDColor](https://www.libreyolo.com/docs/models/ddcolor), [HVI-CIDNet](https://www.libreyolo.com/docs/models/hvi-cidnet), [LaMa](https://www.libreyolo.com/docs/models/lama), [NAFNet](https://www.libreyolo.com/docs/models/nafnet), [QuickSRNet](https://www.libreyolo.com/docs/models/quicksrnet), [Real-ESRGAN](https://www.libreyolo.com/docs/models/real-esrgan), [SwinIR](https://www.libreyolo.com/docs/models/swinir) |
- | **Background removal** | [BiRefNet](https://www.libreyolo.com/docs/models/birefnet), [FeyNobg](https://www.libreyolo.com/docs/models/feynobg), [BEN2](https://www.libreyolo.com/docs/models/ben2), [ViTMatte](https://www.libreyolo.com/docs/models/vitmatte) |
- | **OCR** | [PP-OCR](https://www.libreyolo.com/docs/models/pp-ocrv5), DeepSeek-OCR, DeepSeek-OCR-2, PaddleOCR-VL, GLM-OCR, Falcon OCR |
- | **Document layout** | PP-DocLayoutV3 (regions plus reading order) |
- | **Point detection** | [FOMO](https://www.libreyolo.com/docs/models/fomo), [LocateAnything](https://www.libreyolo.com/docs/models/locate-anything) |
- | **Gaze** | [L2CS](https://www.libreyolo.com/docs/models/l2cs) |
- | **Open vocabulary and VLMs** | [Grounding DINO](https://www.libreyolo.com/docs/models/grounding-dino), [OWLv2](https://www.libreyolo.com/docs/models/owlv2), [OmDet-Turbo](https://www.libreyolo.com/docs/models/omdet-turbo), [OV-DEIM](https://www.libreyolo.com/docs/models/ov-deim), [Florence-2](https://www.libreyolo.com/docs/models/florence-2), [Kosmos-2](https://www.libreyolo.com/docs/models/kosmos-2), [Qwen3-VL](https://www.libreyolo.com/docs/models/qwen3-vl), [InternVL3](https://www.libreyolo.com/docs/models/internvl3), [LFM2-VL](https://www.libreyolo.com/docs/models/lfm2-vl), [North Micro Vision](https://www.libreyolo.com/docs/models/northmicrovision), [SmolVLM2](https://www.libreyolo.com/docs/models/smolvlm2), [Gemma 4](https://www.libreyolo.com/docs/models/gemma-4), [Moondream](https://www.libreyolo.com/docs/models/moondream), [Molmo2](https://www.libreyolo.com/docs/models/molmo2), [WeVisOne-VL](https://www.libreyolo.com/docs/models/wevisone-vl), [MODUS](https://www.libreyolo.com/docs/models/libremodus) |
- | **Robot actions (VLA)** | [SmolVLA](https://www.libreyolo.com/docs/models/librevla), [ACT](https://www.libreyolo.com/docs/models/librevla), [Diffusion Policy](https://www.libreyolo.com/docs/models/librevla) |
- | **Image generation** | [FLUX.2 klein](https://www.libreyolo.com/docs/models/flux2-klein), [SANA-Sprint](https://www.libreyolo.com/docs/models/sana-sprint), PRX, PRXPixel, Speridlabs Iris-Z, Open Vision Banana |
- | **3D generation** | [Speridlabs Idol](https://www.libreyolo.com/docs/models/speridlabs-idol), TripoSplat, TRELLIS, TRELLIS.2, LGM, TripoSR, CRM, Direct3D-S2 |
- </details>
- Per-family sizes, checkpoints and parity evidence live in the
- [model reference](https://www.libreyolo.com/docs/models).
- ## Follow us
- - [LinkedIn](https://www.linkedin.com/company/libreyolo/)
- - [Reddit](https://www.reddit.com/r/LibreYOLO/)
- ## Contributors
- <a href="https://github.com/LibreYOLO/libreyolo/graphs/contributors">
- <img src="https://contrib.rocks/image?repo=LibreYOLO/libreyolo" alt="LibreYOLO contributors">
- </a>
- ## Sponsors
- LibreYOLO is built in free time. To take it to the next level, we want to
- train new models, such as LibreYOLO26 and MIT retrains of families with
- restrictive weights. That costs rented GPU time, and to pay for it we created
- the LibreYOLO Sponsorship Program. You can read all the details in
- [this article](https://www.libreyolo.com/sponsors).
- [](https://opencollective.com/libreyolo)
- **Supporters.** Thank you to everyone who has donated:
- <a href="https://github.com/picjul" title="Mattia Di Giusto"><img src="docs/readme/supporters/picjul.png" width="48" height="48" alt="Mattia Di Giusto"></a> <a href="https://github.com/Magellol" title="Thomas Lefebvre (ma.ge)"><img src="docs/readme/supporters/Magellol.png" width="48" height="48" alt="Thomas Lefebvre"></a>
- ## License
- - **Code:** MIT License.
- - **Weights:** Check the license of the specific Hugging Face repo before you use one commercially. Every LibreYOLO Hugging Face model states its license.
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.