Instructions to use LibreYOLO/LibreRTDETRv2r18 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LibreYOLO
How to use LibreYOLO/LibreRTDETRv2r18 with LibreYOLO:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
LibreRTDETRv2r18
RT-DETRv2 ResNet-18_vd COCO detector, repackaged for the LibreYOLO framework.
Source
Derived from https://github.com/lyuwenyu/RT-DETR at commit a21d516aca15da57e65f35c47659c7535ad2b6b3 and upstream checkpoint:
https://github.com/lyuwenyu/storage/releases/download/v0.1/rtdetrv2_r18vd_120e_coco_rerun_48.1.pth.
The ResNet_vd backbone lineage traces to PaddleClas. Both projects are Apache-2.0 licensed.
Modifications
EMA weights were unwrapped, encoder.input_proj keys were remapped from named submodules to LibreYOLO's sequential layout where required, and the checkpoint was saved with LibreYOLO metadata schema v1.0. Learned tensors are unchanged apart from layout-compatible key names.
Validation
This file was validated locally with LibreYOLO checkpoint metadata schema v1.0, loaded through LibreYOLO, and evaluated on coco128.yaml with workers=0, batch=8, conf=0.001, iou=0.6.
| Metric | Value |
|---|---|
| COCO128 mAP50-95 | 0.6321 |
| COCO128 mAP50 | 0.8128 |
| SHA256 | aa51698d1921e34f4e8d026b14bd916721e96d3ab8d0b1e3072a9e7c2fd620d6 |
| File size | 77.4 MB |
Usage
pip install libreyolo
from libreyolo import LibreYOLO
model = LibreYOLO("LibreRTDETRv2r18.pt")
results = model("image.jpg", save=True)
License
Apache License 2.0. See the LICENSE and NOTICE files in this repository.
- Downloads last month
- 173