Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: onnx
|
| 3 |
+
tags:
|
| 4 |
+
- object-detection
|
| 5 |
+
- yolo
|
| 6 |
+
- yolox
|
| 7 |
+
- yolo9
|
| 8 |
+
- rfdetr
|
| 9 |
+
- onnxruntime-web
|
| 10 |
+
- webgpu
|
| 11 |
+
- browser
|
| 12 |
+
- libreyolo
|
| 13 |
+
license: mit
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
# LibreYOLO Web — Model Zoo
|
| 17 |
+
|
| 18 |
+
Pre-exported ONNX models for [libreyolo-web](https://github.com/xuban-ceccon/libreyolo-web), the browser-based multi-family YOLO detection library.
|
| 19 |
+
|
| 20 |
+
## Usage
|
| 21 |
+
|
| 22 |
+
```bash
|
| 23 |
+
npm install libreyolo-web onnxruntime-web
|
| 24 |
+
```
|
| 25 |
+
|
| 26 |
+
```typescript
|
| 27 |
+
import { loadModel } from 'libreyolo-web';
|
| 28 |
+
|
| 29 |
+
// Auto-downloads from this repo, zero config
|
| 30 |
+
const model = await loadModel('yolox-nano');
|
| 31 |
+
const result = await model.predict(imageElement);
|
| 32 |
+
```
|
| 33 |
+
|
| 34 |
+
## Available Models
|
| 35 |
+
|
| 36 |
+
### YOLOX (letterbox + BGR + 0-255)
|
| 37 |
+
| Name | Input | Size | File |
|
| 38 |
+
|------|-------|------|------|
|
| 39 |
+
| `yolox-nano` | 416 | 3.6MB | `yolox_n.onnx` |
|
| 40 |
+
| `yolox-tiny` | 416 | 19MB | `yolox_t.onnx` |
|
| 41 |
+
| `yolox-small` | 640 | 34MB | `yolox_s.onnx` |
|
| 42 |
+
| `yolox-medium` | 640 | 97MB | `yolox_m.onnx` |
|
| 43 |
+
| `yolox-large` | 640 | 207MB | `yolox_l.onnx` |
|
| 44 |
+
| `yolox-xlarge` | 640 | 378MB | `yolox_x.onnx` |
|
| 45 |
+
|
| 46 |
+
### YOLO9 (resize + RGB + /255)
|
| 47 |
+
| Name | Input | Size | File |
|
| 48 |
+
|------|-------|------|------|
|
| 49 |
+
| `yolo9-tiny` | 640 | 8MB | `yolo9_t.onnx` |
|
| 50 |
+
| `yolo9-small` | 640 | 28MB | `yolo9_s.onnx` |
|
| 51 |
+
| `yolo9-medium` | 640 | 77MB | `yolo9_m.onnx` |
|
| 52 |
+
| `yolo9-compact` | 640 | 97MB | `yolo9_c.onnx` |
|
| 53 |
+
|
| 54 |
+
### RF-DETR (resize + ImageNet norm, no NMS)
|
| 55 |
+
| Name | Input | Size | File |
|
| 56 |
+
|------|-------|------|------|
|
| 57 |
+
| `rfdetr-nano` | 384 | 103MB | `rfdetr_n.onnx` |
|
| 58 |
+
| `rfdetr-small` | 512 | 109MB | `rfdetr_s.onnx` |
|
| 59 |
+
| `rfdetr-medium` | 576 | 115MB | `rfdetr_m.onnx` |
|
| 60 |
+
| `rfdetr-large` | 704 | 116MB | `rfdetr_l.onnx` |
|
| 61 |
+
|
| 62 |
+
## Training
|
| 63 |
+
|
| 64 |
+
All models were trained on COCO using [libreyolo](https://github.com/LibreYOLO/libreyolo) (Python) and exported to ONNX with `model.export(format='onnx', simplify=True)`.
|
| 65 |
+
|
| 66 |
+
## License
|
| 67 |
+
|
| 68 |
+
MIT
|