Update app.py

app.py

@@ -1,56 +1,39 @@
-import logging
 import re
 import gradio as gr
 from PIL import Image
-from transformers import (
-    BlipProcessor,
-    BlipForConditionalGeneration,
-    pipeline,
-)
-
-# Set up logging
-logging.basicConfig(level=logging.INFO)
-
-# 1) BLIP captioner (large model for richer captions)
-caption_processor = BlipProcessor.from_pretrained(
-    "Salesforce/blip-image-captioning-large",
-    use_fast=False
-)
-caption_model = BlipForConditionalGeneration.from_pretrained(
-    "Salesforce/blip-image-captioning-large"
-)
+from transformers import pipeline
 
+# 1) BLIP captioner (rich COCO captions)
 caption_pipe = pipeline(
     task="image-to-text",
-    model=caption_model,
-    processor=caption_processor,
-    device=-1,
+    model="Salesforce/blip-image-captioning-large",
+    device=-1,              # force CPU
     max_length=64,
     do_sample=False,
 )
 
-# 2) Flan-T5 pipelines for category, analysis, suggestions
-FLAN_MODEL = "google/flan-t5-large"
+# 2) Flan-T5 for text‐to‐text
+FLAN = "google/flan-t5-large"
 category_pipe = pipeline(
     "text2text-generation",
-    model=FLAN_MODEL,
-    tokenizer=FLAN_MODEL,
+    model=FLAN,
+    tokenizer=FLAN,
     max_new_tokens=32,
     do_sample=True,
     temperature=1.0,
 )
 analysis_pipe = pipeline(
     "text2text-generation",
-    model=FLAN_MODEL,
-    tokenizer=FLAN_MODEL,
+    model=FLAN,
+    tokenizer=FLAN,
     max_new_tokens=256,
     do_sample=True,
     temperature=1.0,
 )
 suggestion_pipe = pipeline(
     "text2text-generation",
-    model=FLAN_MODEL,
-    tokenizer=FLAN_MODEL,
+    model=FLAN,
+    tokenizer=FLAN,
     max_new_tokens=256,
     do_sample=True,
     temperature=1.0,
@@ -63,79 +46,74 @@ def get_recommendations():
         "https://i.imgur.com/wp3Wzc4.jpeg",
         "https://i.imgur.com/5e2xOA4.jpeg",
         "https://i.imgur.com/txjRk98.jpeg",
-        "https://i.imgur.com/rQ4AYl0.jpeg",
-        "https://i.imgur.com/bDzwD04.jpeg",
-        "https://i.imgur.com/fLMngXI.jpeg",
-        "https://i.imgur.com/nYEJzxt.png",
-        "https://i.imgur.com/Xj92Cjv.jpeg",
     ]
 
 def process(image: Image):
-    # Step 1: Generate BLIP caption
+    # 1) BLIP caption
     caption = caption_pipe(image)[0]["generated_text"].strip()
-    logging.info(f"RAW CAPTION: {caption}")
 
-    # Step 2: Category label
+    # 2) Single‐label category
     cat_prompt = (
-        f"Caption: {caption}\n"
-        "Provide a single concise category label for this ad (e.g. 'Food Ad', 'Fitness Promotion'):"
+        f"Caption: {caption}\n\n"
+        "Give me one concise category label for this ad (e.g. 'Fitness', 'Food'):"
     )
-    raw_cat = category_pipe(cat_prompt)[0]["generated_text"].strip()
-    category = raw_cat.splitlines()[0]
-    logging.info(f"RAW CATEGORY: {raw_cat}")
+    category = category_pipe(cat_prompt)[0]["generated_text"].strip().splitlines()[0]
 
-    # Step 3: Five-sentence Analysis
+    # 3) Five‐sentence analysis
     ana_prompt = (
-        f"Caption: {caption}\n"
+        f"Caption: {caption}\n\n"
         "Write exactly five sentences explaining what this ad communicates and its emotional impact."
     )
     raw_ana = analysis_pipe(ana_prompt)[0]["generated_text"].strip()
-    sentences = re.split(r'(?<=[.!?])\s+', raw_ana)
-    analysis = " ".join(sentences[:5])
-    logging.info(f"RAW ANALYSIS: {raw_ana}")
+    # ensure exactly five sentences
+    sents = re.split(r'(?<=[.!?])\s+', raw_ana)
+    analysis = " ".join(sents[:5])
 
-    # Step 4: Five bullet-point Suggestions
+    # 4) Five bullet‐point suggestions
     sug_prompt = (
-        f"Caption: {caption}\n"
+        f"Caption: {caption}\n\n"
         "Suggest five distinct improvements for this ad. "
-        "Each suggestion must start with '- ' and be one actionable sentence."
+        "Each suggestion must start with '- ' and be one sentence."
     )
     raw_sug = suggestion_pipe(sug_prompt)[0]["generated_text"].strip()
-    lines = [l for l in raw_sug.splitlines() if l.strip().startswith("-")]
-    if len(lines) < 5:
+    bullets = [l for l in raw_sug.splitlines() if l.strip().startswith("-")]
+    # pad/truncate to 5
+    if len(bullets) < 5:
         extras = [l.strip() for l in raw_sug.splitlines() if l.strip()]
         for ex in extras:
-            if len(lines) >= 5:
-                break
-            prefix = "- " if not ex.startswith("-") else ""
-            lines.append(prefix + ex.lstrip("- ").strip())
-    suggestions = "\n".join(lines[:5])
-    logging.info(f"RAW SUGGESTIONS:\n{raw_sug}")
+            if len(bullets) >= 5: break
+            line = ex if ex.startswith("-") else "- " + ex
+            bullets.append(line)
+    suggestions = "\n".join(bullets[:5])
 
     return caption, category, analysis, suggestions, get_recommendations()
 
-with gr.Blocks(theme=gr.themes.Default(primary_hue="blue")) as demo:
+with gr.Blocks() as demo:
     gr.Markdown("## 📢 Smart Ad Analyzer")
     gr.Markdown(
-        "Upload an image ad to get: a **BLIP Caption** (debug), a **Category**, a "
-        "**five-sentence Analysis**, **five bullet-point Suggestions**, and **Example Ads**."
+        "Upload an image ad to get:\n"
+        "- **BLIP Caption** (debug)\n"
+        "- **Category**\n"
+        "- **Five-sentence Analysis**\n"
+        "- **Five bullet-point Suggestions**\n"
+        "- **Example Ads**"
     )
 
     with gr.Row():
-        img       = gr.Image(type="pil", label="Upload Ad Image")
+        inp = gr.Image(type="pil", label="Upload Ad Image")
         with gr.Column():
-            debug_cap = gr.Textbox(label="🔍 BLIP Caption (debug)", interactive=False)
-            cat_out   = gr.Textbox(label="Ad Category", interactive=False)
-            ana_out   = gr.Textbox(label="Ad Analysis", lines=5, interactive=False)
-            sug_out   = gr.Textbox(label="Improvement Suggestions", lines=5, interactive=False)
-            btn       = gr.Button("Analyze Ad")
+            cap_out = gr.Textbox(label="🔍 BLIP Caption", interactive=False)
+            cat_out = gr.Textbox(label="Ad Category", interactive=False)
+            ana_out = gr.Textbox(label="Ad Analysis", lines=5, interactive=False)
+            sug_out = gr.Textbox(label="Improvement Suggestions", lines=5, interactive=False)
+            btn     = gr.Button("Analyze Ad")
 
-    gallery = gr.Gallery(label="Example Ads", show_label=True)
+    gallery = gr.Gallery(label="Example Ads").style(grid=[5], height="auto")
 
     btn.click(
         fn=process,
-        inputs=[img],
-        outputs=[debug_cap, cat_out, ana_out, sug_out, gallery],
+        inputs=[inp],
+        outputs=[cap_out, cat_out, ana_out, sug_out, gallery],
     )
 
     gr.Markdown("Made by Simon Thalmay")