Create app.py

app.py

@@ -0,0 +1,164 @@
+# =========================================================
+# HUGGINGFACE DRIVER STATE DETECTION APP
+# =========================================================
+#
+# Upload driver image
+# → Predict fatigue state
+#
+# Classes:
+# - alert
+# - sleepy
+# - slowBlink
+# - yawning
+#
+# =========================================================
+
+import gradio as gr
+import numpy as np
+import cv2
+
+from tensorflow.keras.models import load_model
+from tensorflow.keras.preprocessing.image import img_to_array
+
+# =========================================================
+# LOAD MODEL
+# =========================================================
+
+model = load_model("final_driver_state_model.h5")
+
+CLASS_NAMES = [
+    "alert",
+    "sleepy",
+    "slowBlink",
+    "yawning"
+]
+
+# =========================================================
+# RISK MAPPING
+# =========================================================
+
+RISK_LEVELS = {
+    "alert": "SAFE",
+    "sleepy": "HIGH RISK",
+    "slowBlink": "MEDIUM RISK",
+    "yawning": "LOW RISK"
+}
+
+# =========================================================
+# PREDICTION FUNCTION
+# =========================================================
+
+def predict_driver_state(image):
+
+    # =============================================
+    # IMAGE PREPROCESSING
+    # =============================================
+
+    image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+
+    image = cv2.resize(image, (224,224))
+
+    image = image.astype("float32") / 255.0
+
+    image = img_to_array(image)
+
+    image = np.expand_dims(image, axis=0)
+
+    # =============================================
+    # MODEL PREDICTION
+    # =============================================
+
+    prediction = model.predict(image, verbose=0)
+
+    class_index = np.argmax(prediction)
+
+    predicted_class = CLASS_NAMES[class_index]
+
+    confidence = float(np.max(prediction))
+
+    risk_level = RISK_LEVELS[predicted_class]
+
+    # =============================================
+    # ALL CLASS PROBABILITIES
+    # =============================================
+
+    confidence_scores = {}
+
+    for i, class_name in enumerate(CLASS_NAMES):
+
+        confidence_scores[class_name] = float(
+            prediction[0][i]
+        )
+
+    # =============================================
+    # RESULT TEXT
+    # =============================================
+
+    result = f"""
+🚗 DRIVER STATE ANALYSIS
+
+Prediction: {predicted_class}
+
+Confidence: {confidence:.2f}
+
+Risk Level: {risk_level}
+"""
+
+    return result, confidence_scores
+
+# =========================================================
+# GRADIO UI
+# =========================================================
+
+title = "🚗 AI Driver Safety Detection System"
+
+description = """
+Upload a driver image to detect fatigue state using Deep Learning.
+
+### Supported Driver States
+- Alert
+- Sleepy
+- Slow Blink
+- Yawning
+
+### AI Features
+✅ CNN-Based Classification  
+✅ Fatigue Risk Analysis  
+✅ Real-Time Prediction Engine  
+"""
+
+# =========================================================
+# INTERFACE
+# =========================================================
+
+interface = gr.Interface(
+
+    fn=predict_driver_state,
+
+    inputs=gr.Image(
+        type="numpy",
+        label="Upload Driver Image"
+    ),
+
+    outputs=[
+        gr.Textbox(
+            label="Prediction Result"
+        ),
+
+        gr.Label(
+            label="Class Confidence Scores"
+        )
+    ],
+
+    title=title,
+
+    description=description,
+
+    theme="soft"
+)
+
+# =========================================================
+# LAUNCH
+# =========================================================
+
+interface.launch()