medieval-data/yolov8-mgh

MGH Critical Edition YOLO Model

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.

Dataset

The model is trained on the dataset available at: medieval-data/mgh-critical-edition-layout. This dataset comprises images from medieval critical editions and their associated annotations.

Training Details

• Architecture: YOLOv8n
• Pretrained Model: yolov8n.pt
• Image Size: 640
• Batch Size: 25
• Augmentation: Enabled
• Epochs: 300

Evaluation Metrics

Forthcoming...

Usage

To utilize this model in your projects, you can use the ultralytics YOLO library. Here's a simple code snippet to get you started:

git clone https://huggingface.co/medieval-data/yolov8-mgh

cd clone yolov8-mgh

from ultralytics import YOLO
import cv2
from matplotlib import pyplot as plt

model = YOLO("yolo8v-mgh.pt")

# Prediction on an image
image_path = "page_103.jpg"
results = model(image_path)

# Visualize the results
results[0].boxes.data.tolist()

# Load the image
image = cv2.imread(image_path)
threshold = 0.5
# Draw bounding boxes on the image
for result in results[0].boxes.data.tolist():
    x1, y1, x2, y2, score, class_id = result
    if score > threshold:
        cv2.rectangle(image, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 4)
        cv2.putText(image, results[0].names[int(class_id)].upper(), (int(x1), int(y1 - 10)),
                    cv2.FONT_HERSHEY_SIMPLEX, 1.3, (0, 255, 0), 3, cv2.LINE_AA)

# Convert BGR image to RGB for plotting
image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

# Display the image in the notebook
plt.figure(figsize=(10, 10))
plt.imshow(image_rgb)
plt.axis('off')
plt.show()

Expected Output