Kansallisarkisto/court-records-textline-detection

Text line detection from Finnish 19th century Court Records

The model is trained to find text lines from digitized 19th century court record documents. The model has been trained using yolov8x-seg by Ultralytics as the base model.

Intended uses & limitations

Most of the training data consist of handwritten documents, but the model appears to generalize quite well also to typeset data.

Training data

Training dataset consisted of 4615 digitized and annotated 19th century court record documents, while validation dataset contained 574 annotated document images.

Training procedure

This model was trained using 2 NVIDIA RTX A6000 GPUs with the following hyperparameters:

• image size: 640
• learning rate (lr0): 0.05
• train batch size: 32
• epochs: 100
• patience: 10 epochs
• optimizer: SGD
• scheduler: cosine learning rate scheduler (cos_lr=True)
• workers: 4

Default settings were used for other training hyperparameters (find more information here).

Model training was performed using the following code:

from ultralytics import YOLO

# Use pretrained Yolo segmentation model
model = YOLO('yolov8x-seg.pt')                                                                                                               

# Path to .yaml file where data location and object classes are defined
yaml_path = 'text_lines.yaml'

# Start model training with the defined parameters
model.train(data=yaml_path, name='model_name', epochs=100, imgsz=640, workers=4, optimizer='SGD', lr0=0.05, seed=551, val=True, cos_lr=True, patience=10, batch=32, device=[0,1])

Evaluation results

Evaluation results using the validation dataset are listed below:

Class	Images	Class instances	Box precision	Box recall	Box mAP50	Box mAP50-95	Mask precision	Mask recall	Mask mAP50	Mask mAP50-95
Text line	574	43156	0.912	0.888	0.949	0.701	0.935	0.907	0.954	0.55

More information on the performance metrics can be found here.

Inference

If the model file tuomiokirja_lines_05122023.pt is downloaded to a folder \models\tuomiokirja_lines_05122023.pt and the input image path is `\data\image.jpg', inference can be perfomed using the following code:

from ultralytics import YOLO

# Initialize model
model = YOLO('\models\tuomiokirja_lines_05122023.pt')
prediction_results = model.predict(source='\data\image.jpg', save=True)

More information for available inference arguments can be found here.