Update README.md

README.md

@@ -14,11 +14,11 @@ metrics:
 library_name: tensorflow
 ---
 
-# 🏥 PCOS Detection with Explainable AI
+#  PCOS Detection with Explainable AI
 
 A deep learning model for **Polycystic Ovary Syndrome (PCOS)** detection from ultrasound images with **Grad-CAM** visualization for clinical interpretability.
 
-## 🎯 Model Overview
+##  Model Overview
 
 - **Architecture**: Dual-path CNN with multi-head attention
 - **Input**: 224×224 RGB ultrasound images  
@@ -65,7 +65,7 @@ CLASS_NAMES = ["infected", "noninfected"]
 # Download model from HF
 # ============================================================
 MODEL_PATH = hf_hub_download(repo_id=HF_MODEL_REPO, filename=MODEL_FILENAME)
-print(f"✅ Model downloaded to: {MODEL_PATH}")
+print(f" Model downloaded to: {MODEL_PATH}")
 
 # ============================================================
 # Custom Lambda Functions
@@ -124,7 +124,7 @@ output = tf.keras.layers.Activation('softmax', name='softmax')(logits)
 
 model = Model(input_layer, output)
 model.load_weights(MODEL_PATH)
-print("✅ Weights loaded successfully")
+print(" Weights loaded successfully")
 
 # ============================================================
 # Load & Preprocess Image
@@ -144,8 +144,8 @@ pred = model.predict(img_array, verbose=0)[0]
 pred_class = np.argmax(pred)
 confidence = pred[pred_class]
 
-print(f"\n🔍 Prediction: {CLASS_NAMES[pred_class]}")
-print(f"📊 Confidence: {confidence:.2%}")
+print(f"\n Prediction: {CLASS_NAMES[pred_class]}")
+print(f" Confidence: {confidence:.2%}")
 
 # ============================================================
 # Grad-CAM
@@ -230,7 +230,7 @@ img_array = np.expand_dims(img, axis=0)
 - **Blue/Cool regions**: Low influence on decision
 - **Dual visualization**: Separate heatmaps for upper and lower ovarian regions
 
-## 📊 Model Architecture
+##  Model Architecture
 
 ```
 Input (224×224×3)
@@ -247,7 +247,7 @@ Input (224×224×3)
 - Multi-head attention for feature fusion
 - Logits-based Grad-CAM (fixes saturated softmax gradients)
 
-## 📈 Dataset
+##  Dataset
 
 - **Total**: 11,784 ultrasound images
 - **PCOS-positive**: 6,784 images (57.5%)
@@ -255,34 +255,34 @@ Input (224×224×3)
 - **Source**: 3 clinics (2018-2022), expert-annotated
 - **Dataset**: [PCOS XAI Ultrasound](https://www.kaggle.com/datasets/ibadeus/pcos-xai-ultrasound-dataset)
 
-## ⚠️ Important Notes
+##  Important Notes
 
 **Clinical Use:**
-- ⚠️ Research purposes only - NOT FDA approved
-- ⚠️ Not a diagnostic tool - requires professional validation
-- ⚠️ Must be validated on local datasets before clinical deployment
+-  Research purposes only - NOT FDA approved
+-  Not a diagnostic tool - requires professional validation
+-  Must be validated on local datasets before clinical deployment
 
 **Technical:**
 - Fixed 224×224 input size required
 - RGB images only
 - Model performance may vary across different ultrasound machines
 
-## 📝 Citation
+##  Citation
 
 ```bibtex
 @misc{pcos_xai_2024,
   title={PCOS Detection with Explainable AI},
   author={Dehsahk-AI},
-  year={2024},
+  year={2025},
   url={https://huggingface.co/Dehsahk-AI/Pcos-Detect}
 }
 ```
 
-## 📜 License
+##  License
 
 MIT License - See LICENSE file for details.
 
-## 🙏 Acknowledgments
+##  Acknowledgments
 
 - Grad-CAM: Selvaraju et al. (ICCV 2017)
 - Multi-head Attention: Vaswani et al. (NeurIPS 2017)