nikolai40/deep-dating

A Generalized Deep Learning-Based Pipeline for Historical Manuscript Dating

Contains all trained models for the dating pipeline

In each directory models are named as follows:

• PyTorch Dating CNN: "model_epoch_[epoch_number]_split_[split_number].pt"
• SVM feature aggregation: "classifier_model_split_[split_number].pkl"

CSV files indicate the performance while training.

Repository is structured as follows:

• Binet_aug_norm: BiNet model with synthetic images
• Binet_norm: Standard BiNet model
• CLAMM_P1_Crossval: 5-fold cross-validation for the CLaMM dataset
• HIMANIS_P1_Crossval: 5-fold cross-validation for the Himanis dataset
• MPS_P1_Crossval: 5-fold cross-validation for the MPS dataset
• SCRIBBLE_P1_Crossval: 5-fold cross-validation for the ScribbleLens dataset