Add pipeline tag and improve model card
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,6 +1,23 @@
|
|
| 1 |
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
datasets:
|
| 4 |
- juliozhao/doclayout-yolo-DocLayNet
|
|
|
|
|
|
|
| 5 |
---
|
| 6 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
|
|
|
| 2 |
datasets:
|
| 3 |
- juliozhao/doclayout-yolo-DocLayNet
|
| 4 |
+
license: apache-2.0
|
| 5 |
+
pipeline_tag: object-detection
|
| 6 |
---
|
| 7 |
+
|
| 8 |
+
# DocLayout-YOLO
|
| 9 |
+
|
| 10 |
+
This model is a real-time document layout detector introduced in the paper [DocLayout-YOLO: Axis-Attentive Document Layout Detector with Mesh-Guided Synthetic Pretraining](https://huggingface.co/papers/2410.12628). It is based on YOLOv10 and is enhanced with diverse synthetic document pretraining and page-axis context modeling. The model can detect various document elements such as text, titles, figures, tables, etc.
|
| 11 |
+
|
| 12 |
+
For usage, please refer to the [GitHub repository](https://github.com/opendatalab/DocLayout-YOLO/tree/release-v2).
|
| 13 |
+
|
| 14 |
+
An example of loading the model:
|
| 15 |
+
|
| 16 |
+
```python
|
| 17 |
+
from doclayout_yolo import YOLOv10
|
| 18 |
+
|
| 19 |
+
model = YOLOv10.from_pretrained("juliozhao/DocLayout-YOLO-DocStructBench")
|
| 20 |
+
|
| 21 |
+
# Perform prediction
|
| 22 |
+
det_res = model.predict("path/to/image", imgsz=1024, conf=0.2, device="cuda:0")
|
| 23 |
+
```
|