Update app.py

app.py

@@ -7,6 +7,7 @@ import requests
 import random
 import dom
 import os
+import time
 
 NUM_IMAGES = 2
 
@@ -23,6 +24,20 @@ headers = {"Authorization": f"Bearer {os.getenv('api_token')}"}
 model_id_image_description = "vikhyatk/moondream2"
 revision = "2024-08-26"
 
+# Para medir el rendimiento de los métodos, voy a crear este decorador, que simplemente imprime en nuestra terminal
+#  el tiempo de ejecucion de los metodos que tengan los modelos y los usen, de esta manera podremos estudiar 
+# el tiempo que este cada modelo activo
+def measure_performance(func):
+    def wrapper(*args, **kwargs):
+        print(f"Starting execution of '{func.__name__}' with 'args={args}, kwargs={kwargs}'")
+        start = time.time()
+        result = func(*args, **kwargs)
+        end = time.time()
+        duration = end - start
+        print(f"Execution time of '{func.__name__}' with 'args={args}, kwargs={kwargs}': {duration:.4f} seconds")
+        return result
+    return wrapper
+
 torch_dtype = torch.float32
 if torch.cuda.is_available():
     torch_dtype = torch.bfloat16  # Optimización en GPU
@@ -32,6 +47,7 @@ print("Cargando modelo de descripción de imágenes...")
 model_description = AutoModelForCausalLM.from_pretrained(model_id_image_description, trust_remote_code=True, revision=revision)
 tokenizer_description = AutoTokenizer.from_pretrained(model_id_image_description, revision=revision)
 
+@measure_performance
 def generate_description(image_path):
     image_test = Image.open(image_path)
     enc_image = model_description.encode_image(image_test)
@@ -42,6 +58,7 @@ def query(payload):
 	response = requests.post(API_URL, headers=headers, json=payload)
 	return response.content
 
+@measure_performance
 def generate_image_by_description(description, avatar_style=None):
     images = []
     for _ in range(NUM_IMAGES):
@@ -52,29 +69,34 @@ def generate_image_by_description(description, avatar_style=None):
         image_bytes = query({"inputs": prompt, "parameters": {"seed": random.randint(0, 1000)}})
         image = Image.open(io.BytesIO(image_bytes))
         images.append(image)
-    print(images)
     return images
 
 def process_and_generate(image, avatar_style):
     description = generate_description(image)
     return generate_image_by_description(description, avatar_style)
 
-with gr.Blocks(js=dom.generate_title) as demo:
-    with gr.Row(): 
-        gr.Markdown(dom.generate_markdown)
-        gr.Markdown(dom.models)
+with gr.Blocks(js=dom.generate_title) as demo:        
     with gr.Row():
         with gr.Column(scale=2, min_width=300):
             selected_image = gr.Image(type="filepath", label="Upload an Image of the Pigeon", height=300)
             example_image = gr.Examples(["./examples/pigeon.webp"], label="Example Images", inputs=[selected_image])
             avatar_style = gr.Radio(
                 ["Realistic", "Pixel Art", "Imaginative", "Cartoon"], 
-                label="(optional) Select the avatar style:"
+                label="(optional) Select the avatar style:",
+                value="Pixel Art"
             )
             generate_button = gr.Button("Generate Avatar", variant="primary")
         with gr.Column(scale=2, min_width=300):
             generated_image = gr.Gallery(type="pil", label="Generated Avatar", height=300)
             
         generate_button.click(process_and_generate, inputs=[selected_image, avatar_style], outputs=generated_image)
+    with gr.Tab(label="Description"):
+        gr.Markdown(dom.generate_markdown)
+        gr.Markdown(dom.models)
+    with gr.Tab(label="Documentation"):
+        gr.Markdown(dom.doccumentation)
+
+        
+        
 
 demo.launch()