Update app.py

app.py

@@ -1,110 +1,62 @@
-# ================================
-# ENVIRONMENT FIXES (MUST BE FIRST)
-# ================================
 import os
-os.environ["KERAS_BACKEND"] = "tensorflow"
 os.environ["TF_ENABLE_ONEDNN_OPTS"] = "0"
 
-# ================================
-# IMPORTS
-# ================================
 import gradio as gr
 import tensorflow as tf
 import numpy as np
 import pickle
 from PIL import Image
+import json
 
-# ================================
-# LOAD FUSION MODEL (KERAS 3 SAFE)
-# ================================
-fusion_model = tf.keras.models.load_model(
-    "fusion_ticket_model_final.keras",
-    compile=False
-)
+# ----------------------------
+# Rebuild model safely (Keras 3 compatible)
+# ----------------------------
+with open("fusion_config.json", "r") as f:
+    model_json = f.read()
+
+fusion_model = tf.keras.models.model_from_json(model_json)
+fusion_model.load_weights("fusion_weights.weights.h5")
 
-print("✅ Fusion model loaded successfully")
+print("✅ Model rebuilt + weights loaded")
 
-# ================================
-# LOAD TOKENIZER
-# ================================
+# ----------------------------
+# Load tokenizer
+# ----------------------------
 with open("tokenizer.pkl", "rb") as f:
     tokenizer = pickle.load(f)
 
-print("✅ Tokenizer loaded")
-
-# ================================
-# CONSTANTS (MUST MATCH TRAINING)
-# ================================
+# ----------------------------
 IMG_SIZE = (128, 128)
 MAX_LEN = 50
 LABELS = ["Critical", "High", "Medium", "Low"]
 
-# ================================
-# IMAGE PREPROCESSING
-# ================================
-def preprocess_image(image: Image.Image):
-    image = image.convert("RGB")
-    image = image.resize(IMG_SIZE)
-    img = np.asarray(image, dtype=np.float32) / 255.0
-    img = np.expand_dims(img, axis=0)
-    return img
+def preprocess_image(image):
+    image = image.convert("RGB").resize(IMG_SIZE)
+    img = np.array(image, dtype=np.float32) / 255.0
+    return np.expand_dims(img, 0)
 
-# ================================
-# TEXT PREPROCESSING
-# ================================
-def preprocess_text(text: str):
-    if not text:
-        text = ""
-    seq = tokenizer.texts_to_sequences([text])
-    padded = tf.keras.preprocessing.sequence.pad_sequences(
-        seq, maxlen=MAX_LEN
-    )
-    return padded
+def preprocess_text(text):
+    seq = tokenizer.texts_to_sequences([text or ""])
+    return tf.keras.preprocessing.sequence.pad_sequences(seq, maxlen=MAX_LEN)
 
-# ================================
-# PREDICTION FUNCTION (PURE MODEL)
-# ================================
 def predict_ticket(image, text):
     if image is None:
-        return {
-            "Critical": 0.0,
-            "High": 0.0,
-            "Medium": 0.0,
-            "Low": 0.0
-        }
+        return {}
 
     img = preprocess_image(image)
     txt = preprocess_text(text)
 
     probs = fusion_model.predict([img, txt], verbose=0)[0]
+    return dict(zip(LABELS, map(float, probs)))
 
-    return {
-        "Critical": float(probs[0]),
-        "High": float(probs[1]),
-        "Medium": float(probs[2]),
-        "Low": float(probs[3]),
-    }
-
-# ================================
-# GRADIO UI
-# ================================
 interface = gr.Interface(
     fn=predict_ticket,
     inputs=[
-        gr.Image(type="pil", label="📤 Upload Ticket Screenshot"),
-        gr.Textbox(
-            lines=4,
-            placeholder="Describe the issue (recommended)",
-            label="✍️ Ticket Description"
-        )
+        gr.Image(type="pil"),
+        gr.Textbox(lines=3)
     ],
-    outputs=gr.Label(num_top_classes=4, label="🚨 Predicted Severity"),
-    title="🎫 Ticket Severity Classification System",
-    description=(
-        "This system uses a **CNN + NLP Fusion Model** to predict "
-        "ticket urgency from a screenshot and description.\n\n"
-        "**Classes:** Critical | High | Medium | Low"
-    )
+    outputs=gr.Label(num_top_classes=4),
+    title="Ticket Severity Classifier"
 )
 
 interface.launch()