Rename model card to Platrix; add img-test sample photos and cross-links
Browse files- README.md +8 -2
- img-test/sample-01.jpg +0 -0
- img-test/sample-02.jpg +0 -0
- img-test/sample-03.jpg +0 -0
- img-test/sample-04.jpg +0 -0
- img-test/sample-05.jpg +0 -0
- img-test/sample-06.jpg +0 -0
- img-test/sample-07.jpg +0 -0
- img-test/sample-08.jpg +0 -0
- img-test/sample-09.jpg +0 -0
- img-test/sample-10.jpg +0 -0
README.md
CHANGED
|
@@ -18,12 +18,15 @@ tags:
|
|
| 18 |
- object-detection
|
| 19 |
---
|
| 20 |
|
| 21 |
-
#
|
| 22 |
|
| 23 |
Production ONNX models for detecting and reading **Iranian vehicle license
|
| 24 |
plates**, powering the [**Platrix**](https://github.com/AliAkrami1375/Platrix)
|
| 25 |
real-time, self-hosted plate-surveillance system.
|
| 26 |
|
|
|
|
|
|
|
|
|
|
| 27 |
The pipeline is **two-stage**: a YOLO **detector** locates the plate in the
|
| 28 |
frame, then a segmentation-free **CRNN + CTC reader** reads the whole plate at
|
| 29 |
once and returns the standard Iranian layout `DD L DDD DD`
|
|
@@ -68,10 +71,13 @@ instead of shifting the input distribution.
|
|
| 68 |
|
| 69 |
```bash
|
| 70 |
pip install onnxruntime opencv-python-headless numpy
|
| 71 |
-
huggingface-cli download Dibachain/
|
| 72 |
plate_yolo.onnx ocr_crnn.onnx ocr_crnn.labels.json --local-dir models/
|
| 73 |
```
|
| 74 |
|
|
|
|
|
|
|
|
|
|
| 75 |
```python
|
| 76 |
import json, cv2, numpy as np, onnxruntime as ort
|
| 77 |
|
|
|
|
| 18 |
- object-detection
|
| 19 |
---
|
| 20 |
|
| 21 |
+
# Platrix — Iranian License-Plate Recognition Models
|
| 22 |
|
| 23 |
Production ONNX models for detecting and reading **Iranian vehicle license
|
| 24 |
plates**, powering the [**Platrix**](https://github.com/AliAkrami1375/Platrix)
|
| 25 |
real-time, self-hosted plate-surveillance system.
|
| 26 |
|
| 27 |
+
> 📦 **Code & full pipeline:** https://github.com/AliAkrami1375/Platrix
|
| 28 |
+
> 🤗 **This model repo:** https://huggingface.co/Dibachain/Platrix
|
| 29 |
+
|
| 30 |
The pipeline is **two-stage**: a YOLO **detector** locates the plate in the
|
| 31 |
frame, then a segmentation-free **CRNN + CTC reader** reads the whole plate at
|
| 32 |
once and returns the standard Iranian layout `DD L DDD DD`
|
|
|
|
| 71 |
|
| 72 |
```bash
|
| 73 |
pip install onnxruntime opencv-python-headless numpy
|
| 74 |
+
huggingface-cli download Dibachain/Platrix \
|
| 75 |
plate_yolo.onnx ocr_crnn.onnx ocr_crnn.labels.json --local-dir models/
|
| 76 |
```
|
| 77 |
|
| 78 |
+
A few **real test photos** ship in this repo under [`img-test/`](./tree/main/img-test)
|
| 79 |
+
so you can try recognition right away.
|
| 80 |
+
|
| 81 |
```python
|
| 82 |
import json, cv2, numpy as np, onnxruntime as ort
|
| 83 |
|
img-test/sample-01.jpg
ADDED
|
img-test/sample-02.jpg
ADDED
|
img-test/sample-03.jpg
ADDED
|
img-test/sample-04.jpg
ADDED
|
img-test/sample-05.jpg
ADDED
|
img-test/sample-06.jpg
ADDED
|
img-test/sample-07.jpg
ADDED
|
img-test/sample-08.jpg
ADDED
|
img-test/sample-09.jpg
ADDED
|
img-test/sample-10.jpg
ADDED
|