William Mattingly
commited on
Commit
•
9cb5082
1
Parent(s):
7a04879
init
Browse files- README.md +71 -0
- demo.ipynb +0 -0
- output.JPG +0 -0
- page_103.jpg +0 -0
- YOLOv8-mgh.pt → yolo8v-mgh.pt +2 -2
README.md
ADDED
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
## MGH Critical Edition YOLO Model
|
2 |
+
|
3 |
+
This repository hosts a YOLO model specifically trained to detect and annotate various elements from medieval manuscripts. The model is built and trained using the Ultralytics YOLOv8n architecture.
|
4 |
+
|
5 |
+
### Dataset
|
6 |
+
|
7 |
+
The model is trained on the dataset available at: [medieval-data/mgh-critical-edition-layout](https://huggingface.co/datasets/medieval-data/mgh-critical-edition-layout). This dataset comprises images from medieval manuscripts and their associated annotations.
|
8 |
+
|
9 |
+
### Training Details
|
10 |
+
|
11 |
+
- Architecture: YOLOv8n
|
12 |
+
- Pretrained Model: `yolov8n.pt`
|
13 |
+
- Image Size: 640
|
14 |
+
- Batch Size: 25
|
15 |
+
- Augmentation: Enabled
|
16 |
+
- Epochs: 300
|
17 |
+
|
18 |
+
### Evaluation Metrics
|
19 |
+
|
20 |
+
Forthcoming...
|
21 |
+
|
22 |
+
### Usage
|
23 |
+
|
24 |
+
To utilize this model in your projects, you can use the `ultralytics` YOLO library. Here's a simple code snippet to get you started:
|
25 |
+
|
26 |
+
```bash
|
27 |
+
git clone https://huggingface.co/medieval-data/yolov8-mgh
|
28 |
+
```
|
29 |
+
|
30 |
+
```bash
|
31 |
+
cd clone yolov8-mgh
|
32 |
+
```
|
33 |
+
|
34 |
+
```python
|
35 |
+
from ultralytics import YOLO
|
36 |
+
import cv2
|
37 |
+
from matplotlib import pyplot as plt
|
38 |
+
|
39 |
+
model = YOLO("yolo8v-mgh.pt")
|
40 |
+
|
41 |
+
# Prediction on an image
|
42 |
+
image_path = "page_103.jpg"
|
43 |
+
results = model(image_path)
|
44 |
+
|
45 |
+
# Visualize the results
|
46 |
+
results[0].boxes.data.tolist()
|
47 |
+
|
48 |
+
# Load the image
|
49 |
+
image = cv2.imread(image_path)
|
50 |
+
threshold = 0.5
|
51 |
+
# Draw bounding boxes on the image
|
52 |
+
for result in results[0].boxes.data.tolist():
|
53 |
+
x1, y1, x2, y2, score, class_id = result
|
54 |
+
if score > threshold:
|
55 |
+
cv2.rectangle(image, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 4)
|
56 |
+
cv2.putText(image, results[0].names[int(class_id)].upper(), (int(x1), int(y1 - 10)),
|
57 |
+
cv2.FONT_HERSHEY_SIMPLEX, 1.3, (0, 255, 0), 3, cv2.LINE_AA)
|
58 |
+
|
59 |
+
# Convert BGR image to RGB for plotting
|
60 |
+
image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
|
61 |
+
|
62 |
+
# Display the image in the notebook
|
63 |
+
plt.figure(figsize=(10, 10))
|
64 |
+
plt.imshow(image_rgb)
|
65 |
+
plt.axis('off')
|
66 |
+
plt.show()
|
67 |
+
```
|
68 |
+
|
69 |
+
### Expected Output
|
70 |
+
|
71 |
+
![outout](output.JPG)
|
demo.ipynb
ADDED
The diff for this file is too large to render.
See raw diff
|
|
output.JPG
ADDED
|
page_103.jpg
ADDED
![]() |
YOLOv8-mgh.pt → yolo8v-mgh.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ca3fe79e74ac029eb0d23710b687dc56d8afe50cdbc36ec4a358e99cd13b1edf
|
3 |
+
size 6246169
|