Update app.py

app.py

@@ -1,18 +1,18 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
-import os
-import tempfile
+import base64
+import io
 
 # Initialize the Hugging Face Inference Client
 client = InferenceClient()
 
 # Function to analyze plant images
 def analyze_plant_image(image):
-    # Save image to a temporary location with appropriate extension
-    temp_dir = tempfile.mkdtemp()
-    temp_image_path = os.path.join(temp_dir, "uploaded_image.jpg")
-    image.save(temp_image_path, format="JPEG")
-    image_url = f"file://{temp_image_path}"
+    buffered = io.BytesIO()
+    image.save(buffered, format="JPEG")
+    image_base64 = base64.b64encode(buffered.getvalue()).decode("utf-8")
+    image_url = f"data:image/jpeg;base64,{image_base64}"
+
 
     # Create the message structure
     messages = [
@@ -34,15 +34,18 @@ def analyze_plant_image(image):
     ]
 
     # Create the completion request
-    completion = client.chat_completions.create(
+    stream = client.chat.completions.create(
         model="Qwen/Qwen2-VL-7B-Instruct",
         messages=messages,
-        max_tokens=1024
+        max_tokens=1024,
+        stream=True
     )
 
-    # Extract the response
-    description = completion.choices[0].message["content"]
-    return description
+    # Stream content as it is generated
+    output_text = ""
+    for chunk in stream:
+        output_text += chunk.choices[0].delta.content
+        yield output_text
 
 # Create Gradio interface
 with gr.Blocks() as app:
@@ -63,4 +66,4 @@ with gr.Blocks() as app:
     analyze_button.click(fn=analyze_plant_image, inputs=image_input, outputs=output_markdown)
 
 # Run the Gradio app
-app.launch(debug=True)
+app.launch()