Update app.py

app.py

@@ -1,14 +1,14 @@
 import gradio as gr
 from PIL import Image
-import os
 import io
+import os
 from huggingface_hub import InferenceClient
 
-HF_TOKEN = os.environ.get("HF_TOKEN")  # For Hugging Face Spaces
-
+# You may need to set your Hugging Face token as an environment variable:
+# os.environ["HF_TOKEN"] = "your-hf-token"
 client = InferenceClient(
     model="google/gemma-3-4b-it",
-    token=HF_TOKEN
+    token=os.environ.get("HF_TOKEN")  # Add your token if needed for Spaces
 )
 
 def get_recommendations():
@@ -27,27 +27,33 @@ def get_recommendations():
     ]
 
 def gemma_image_analysis(image: Image.Image):
-    # Convert image to bytes and upload as a file
     buf = io.BytesIO()
     image.save(buf, format="PNG")
     buf.seek(0)
     img_bytes = buf.read()
 
-    # Upload image to Hugging Face hub and get a URL
-    img_url = client.upload_image(img_bytes)
-    # Compose multimodal message for Gemma-3
     messages = [
         {
             "role": "system",
             "content": [
-                {"type": "text", "text": "You are an expert ad analyst AI. Analyze the given ad and provide answers for three sections: 1. Category (one word) 2. Analysis (five sentences) 3. Five unique actionable improvement suggestions as a list starting with '- ' each. Output must be in three sections with clear headings."}
+                {"type": "text", "text": (
+                    "You are an expert ad analyst AI. Analyze the given ad and provide answers for three sections:"
+                    " 1. Category (one word),"
+                    " 2. Analysis (five sentences),"
+                    " 3. Five unique, actionable improvement suggestions as a list starting with '- '."
+                    " Output must be in three sections with clear headings."
+                )}
             ]
         },
         {
             "role": "user",
             "content": [
-                {"type": "image_url", "image_url": {"url": img_url}},
-                {"type": "text", "text": "Please respond with:\n\nCategory:\n[category]\n\nAnalysis:\n[5 sentences]\n\nImprovement Suggestions:\n- [suggestion 1]\n- [suggestion 2]\n- [suggestion 3]\n- [suggestion 4]\n- [suggestion 5]\n\nEach suggestion must be unique and actionable."}
+                {"type": "image", "image": img_bytes},
+                {"type": "text", "text": (
+                    "Please respond with:\n\nCategory:\n[category]\n\nAnalysis:\n[5 sentences]\n\nImprovement Suggestions:\n"
+                    "- [suggestion 1]\n- [suggestion 2]\n- [suggestion 3]\n- [suggestion 4]\n- [suggestion 5]\n\n"
+                    "Each suggestion must be unique and actionable. Do not repeat suggestions. If you don't know, say 'not detected'."
+                )}
             ]
         }
     ]
@@ -55,40 +61,55 @@ def gemma_image_analysis(image: Image.Image):
     response = client.chat.completions.create(
         model="google/gemma-3-4b-it",
         messages=messages,
-        max_tokens=500
+        max_tokens=512,
     )
 
     return response.choices[0].message["content"]
 
-def process(image):
+def process(image: Image.Image):
     if image is None:
         return "", "", "", get_recommendations()
 
+    # Run multimodal analysis
     full_output = gemma_image_analysis(image)
-    # Parse the response into 3 sections
-    cat, ana, sugs = "", "", ""
+
+    # Parse the output into sections (simple but robust splitting)
+    category = ""
+    analysis = ""
+    suggestions = ""
+    # Find the sections in model output (robust to little formatting errors)
     try:
-        parts = full_output.split("Analysis:")
-        if len(parts) >= 2:
-            cat = parts[0].replace("Category:", "").strip()
-            rest = parts[1].split("Improvement Suggestions:")
-            if len(rest) == 2:
-                ana = rest[0].strip()
-                sugs = rest[1].strip()
-            else:
-                ana = parts[1].strip()
+        lower = full_output.lower()
+        cat_idx = lower.find("category")
+        ana_idx = lower.find("analysis")
+        sug_idx = lower.find("improvement suggestions")
+
+        if cat_idx != -1 and ana_idx != -1:
+            category = full_output[cat_idx + 8 : ana_idx].strip().strip(":")
+        if ana_idx != -1 and sug_idx != -1:
+            analysis = full_output[ana_idx + 8 : sug_idx].strip().strip(":")
+        if sug_idx != -1:
+            suggestions = full_output[sug_idx + len("improvement suggestions"):].strip()
     except Exception:
-        cat, ana, sugs = "", "", full_output.strip()
+        # Fallback if parsing fails
+        return "", "Analysis parsing failed", "Suggestion parsing failed", get_recommendations()
 
-    return cat, ana, sugs, get_recommendations()
+    return category.strip(), analysis.strip(), suggestions.strip(), get_recommendations()
 
 def main():
-    with gr.Blocks(title="Smart Ad Analyzer (Gemma-3 Edition)") as demo:
-        gr.Markdown("## 📢 Smart Ad Analyzer (Gemma-3 Edition)")
+    with gr.Blocks(title="Smart Ad Analyzer") as demo:
+        gr.Markdown("## 📢 Smart Ad Analyzer (Gemma-3 Model)")
         gr.Markdown(
             """
-            Upload your ad image below and instantly get expert feedback.  
-            Category, analysis, improvement suggestions—and example ads for inspiration.
+            **Upload your ad image below and instantly get expert feedback.**
+
+            This AI tool will analyze your ad and provide:
+            - 📂 **Category** — What type of ad is this?
+            - 📊 **In-depth Analysis** — Five detailed sentences covering message, visuals, emotional impact, and more.
+            - 🚀 **Improvement Suggestions** — Five actionable, unique ways to make your ad better.
+            - 📸 **Inspiration Gallery** — See other effective ads for ideas.
+
+            Perfect for marketers, founders, designers, and anyone looking to boost ad performance with actionable insights!
             """
         )
         with gr.Row():
@@ -110,3 +131,4 @@ def main():
 if __name__ == "__main__":
     demo = main()
     demo.launch()
+