dinov2_Liveness_detection_v2.2.3

This model is a fine-tuned version of nguyenkhoa/dinov2_Liveness_detection_v2.2.2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0224
• Accuracy: 0.9932
• F1: 0.9932
• Recall: 0.9932
• Precision: 0.9933

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 768
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Recall	Precision
0.0562	0.5872	64	0.0385	0.9860	0.9860	0.9860	0.9860
0.0328	1.1743	128	0.0350	0.9884	0.9884	0.9884	0.9885
0.0251	1.7615	192	0.0311	0.9879	0.9879	0.9879	0.9879
0.0185	2.3486	256	0.0296	0.9895	0.9895	0.9895	0.9895
0.0166	2.9358	320	0.0328	0.9897	0.9897	0.9897	0.9898
0.0109	3.5229	384	0.0336	0.9906	0.9906	0.9906	0.9907
0.0098	4.1101	448	0.0249	0.9917	0.9917	0.9917	0.9917
0.0069	4.6972	512	0.0224	0.9932	0.9932	0.9932	0.9933

Evaluate results

• APCER: 0.1827
• BPCER: 0.0089
• ACER: 0.0958
• Accuracy: 0.8700
• F1: 0.8975
• Recall: 0.9911
• Precision: 0.7026

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0