Update app.py

app.py

@@ -2,6 +2,7 @@ import gradio as gr
 import tensorflow as tf
 from transformers import GPT2LMHeadModel, GPT2Tokenizer
 import numpy as np
+from tensorflow.image import resize
 
 # Load the image classification model
 image_classification_model = tf.keras.models.load_model("klasifikasi_pt1.h5")
@@ -11,13 +12,18 @@ gpt2_model_name = "diamantrsd/cerpen-generator-v3"
 gpt2_model = GPT2LMHeadModel.from_pretrained(gpt2_model_name)
 gpt2_tokenizer = GPT2Tokenizer.from_pretrained(gpt2_model_name)
 
+
+
 def classify_and_generate_text(image):
     try:
         # Convert Gradio Image interface output to a NumPy array
         img_array = image.astype('float32') / 255.0
 
-        # Classify the image using the image classification model
-        class_label = image_classification_model.predict(np.expand_dims(img_array, axis=0))
+        # Resize the image to the expected shape (224, 224)
+        img_array_resized = resize(img_array, (224, 224))
+
+        # Classify the resized image using the image classification model
+        class_label = image_classification_model.predict(np.expand_dims(img_array_resized, axis=0))
 
         # Map class label to corresponding category (adjust as needed)
         category = map_class_label_to_category(class_label)
@@ -29,6 +35,7 @@ def classify_and_generate_text(image):
     except Exception as e:
         return f"Error: {str(e)}"
 
+
 def map_class_label_to_category(class_label):
     # Map the class label to a category (replace with your own mapping)
     categories = ['Blazer', 'Blouse', 'Cardigan', 'Dress', 'Jacket',
@@ -41,7 +48,7 @@ def generate_text_with_gpt2(product_category):
     input_ids = gpt2_tokenizer.encode(prompt, return_tensors="pt")
 
     # Adjust parameters as needed
-    max_length = 1000
+    max_length = 50
     no_repeat_ngram_size = 3
     top_k = 50
     top_p = 0.95