BDRC/gyuyig-tsugdri-binary-script-classifier

Gyuyig vs Tsugdri Binary Script Classifier (DINOv3 ViT-S)

Fine-tuned DINOv3 ViT-S for parent script classification:

Gyuyig, Tsugdri

Experiment: dinov3_gyuyig_tsugdri_sub_warmstart (gyuyig_tsugdri_binary_classification)
Pooling: ViT CLS token (last_hidden_state[:, 0, :])
Weights: final_model.pt (best validation macro-F1 across stages A/B/C)

Warm-start: BDRC/4-class-balanced-script-classifier (final_model.pt — prior test acc 82.6%, macro-F1 0.833)

Data

Split	Source
Train / val / test	BDRC/gyuyig-tsugdri-binary-balanced-script-classification-dataset

Test split: balanced benchmark (60 images per parent class, held out of training).

Preprocessing

Split	Mode	Size
train	resize_letterbox	448
val	resize_letterbox	448
test	resize_letterbox	448

Validation metrics (n=60)

Metric	Value
Accuracy	85.0%
Macro F1	0.850
Weighted F1	0.850
AUC-ROC	0.902
Loss	0.4520

Best checkpoint: best_stage_c_last_blocks.pt epoch 7 val macro-F1 0.850

Per-class (validation)

precision    recall  f1-score   support

      Gyuyig       0.89      0.80      0.84        30
     Tsugdri       0.82      0.90      0.86        30

    accuracy                           0.85        60
   macro avg       0.85      0.85      0.85        60
weighted avg       0.85      0.85      0.85        60

Test / benchmark metrics (n=120)

Metric	Value
Accuracy	80.8%
Macro F1	0.808
Weighted F1	0.808
AUC-ROC	0.868
Loss	0.5354

Per-class (test)

precision    recall  f1-score   support

      Gyuyig       0.79      0.83      0.81        60
     Tsugdri       0.82      0.78      0.80        60

    accuracy                           0.81       120
   macro avg       0.81      0.81      0.81       120
weighted avg       0.81      0.81      0.81       120

Training

Stage	Epochs	LR head	LR backbone	Unfrozen blocks
A	7	0.0005	—	0
B	10	0.0001	1e-05	4
C	12	5e-05	1.5e-05	8

Setting	Value
Scheduler	cosine_warmup
Class weights	custom
Label smoothing	0.05
Dropout	0.1

Confusion matrix (test)

True \ Pred	Gyuyig	Tsugdri
Gyuyig	50	10
Tsugdri	13	47

Files

File	Description
final_model.pt	Best val-F1 weights + label maps
results.json	Full metrics, history, warm-start info
config.yaml	Training config
model_card.json	Summary metadata
confusion_matrix.json / .png	Test CM
training_history.png	Stage loss / val F1 curves
split_stats.json / .md	Per-class split counts
inference.py	Classify image paths
requirements-inference.txt	Pip deps

Inference

pip install -r requirements-inference.txt
python inference.py --checkpoint final_model.pt --image path/to/page.jpg --preprocess resize_letterbox --preprocess-size 448

Reproduce training

python experiments/gyuyig-tsugdri/train.py --config experiments/gyuyig-tsugdri/config_warmstart.yaml

Model repo: BDRC/gyuyig-tsugdri-binary-script-classifier