Update README.md

README.md

@@ -1,3 +1,4 @@
+---
 language:
   - en
 library_name: timm
@@ -80,7 +81,7 @@ The model achieves strong discriminative performance with an AUC ROC of 0.9561 o
 Source code and training pipeline are available at:  
 https://github.com/ArchitRastogi20/vit-spoof-detection-pda
 
----
+
 
 ## Model Summary
 
@@ -92,7 +93,7 @@ https://github.com/ArchitRastogi20/vit-spoof-detection-pda
 - Framework: PyTorch with timm
 - Pretraining: ImageNet
 
----
+
 
 ## Intended Use
 
@@ -101,7 +102,7 @@ Potential application domains include biometric authentication systems, access c
 
 The model is not intended for deployment in high risk security environments without additional validation, calibration, and fairness analysis.
 
----
+
 
 ## Dataset
 
@@ -115,7 +116,7 @@ CelebA Spoof is a large scale face anti spoofing dataset containing diverse spoo
 Dataset reference:  
 https://github.com/Davidzhangyuanhan/CelebA-Spoof
 
----
+
 
 ## Data Augmentation
 
@@ -139,7 +140,7 @@ Applied transformations include:
 
 Normalization follows ImageNet statistics used by ViT models.
 
----
+
 
 ## Model Architecture
 
@@ -159,7 +160,7 @@ Key configuration details:
 * Dropout: 0.1
 * Mixed precision training enabled
 
----
+
 
 ## Training Procedure
 
@@ -180,7 +181,7 @@ Training configuration:
 | Early stopping | Enabled                            |
 | Device         | NVIDIA RTX A5000                   |
 
----
+
 
 ## Evaluation
 
@@ -189,7 +190,7 @@ Threshold optimization was applied to balance false acceptance and false rejecti
 
 Reported metrics include accuracy, F1 score, AUC ROC, precision, recall, specificity, FAR, FRR, and EER.
 
----
+
 
 ## Results
 
@@ -218,7 +219,7 @@ Reported metrics include accuracy, F1 score, AUC ROC, precision, recall, specifi
 | FRR    | 0.0232 |
 | EER    | 0.1083 |
 
----
+
 
 ## Confusion Matrix
 
@@ -227,7 +228,7 @@ Reported metrics include accuracy, F1 score, AUC ROC, precision, recall, specifi
 | Actual Spoof | 404             | 267            |
 | Actual Live  | 25              | 1051           |
 
----
+
 
 ## Limitations
 
@@ -235,7 +236,7 @@ Reported metrics include accuracy, F1 score, AUC ROC, precision, recall, specifi
 * High FAR indicates sensitivity to certain spoof patterns
 * No cross dataset evaluation included
 
----
+
 
 ## Citation