final deploy

app.py

@@ -2,98 +2,130 @@ import gradio as gr
 from inference import model_fn, predict_fn
 import os
 
+# Emoji mappings
+EMOTION_EMOJI = {
+    'anger': '😡',
+    'disgust': '🤢',
+    'fear': '😨',
+    'joy': '😄',
+    'neutral': '😐',
+    'sadness': '😢',
+    'surprise': '😲'
+}
+SENTIMENT_EMOJI = {
+    'negative': '😡',
+    'neutral': '😐',
+    'positive': '😄'
+}
+
 # Load model once
 model_dict = model_fn('.')
 
+# Utility to compute top emotion/sentiment across utterances
+def compute_overall(utterances):
+    # aggregate confidences
+    emo_scores = {}
+    senti_scores = {}
+    for utt in utterances:
+        for emo in utt.get('emotions', []):
+            emo_scores.setdefault(emo['label'], []).append(emo['confidence'])
+        for sent in utt.get('sentiments', []):
+            senti_scores.setdefault(sent['label'], []).append(sent['confidence'])
+    # compute average
+    def avg(scores): return sum(scores)/len(scores) if scores else 0
+    emo_avg = [(label, avg(scores)) for label, scores in emo_scores.items()]
+    senti_avg = [(label, avg(scores)) for label, scores in senti_scores.items()]
+    if not emo_avg or not senti_avg:
+        return None
+    # pick top
+    top_emo_label, top_emo_score = max(emo_avg, key=lambda x: x[1])
+    top_sent_label, top_sent_score = max(senti_avg, key=lambda x: x[1])
+    return {
+        'emotion_label': top_emo_label,
+        'emotion_score': top_emo_score,
+        'emotion_emoji': EMOTION_EMOJI.get(top_emo_label, ''),
+        'sentiment_label': top_sent_label,
+        'sentiment_score': top_sent_score,
+        'sentiment_emoji': SENTIMENT_EMOJI.get(top_sent_label, '')
+    }
+
+
 def predict(video):
     temp_path = "temp.mp4"
     try:
-        # Save upload to temp
+        # save upload
         if isinstance(video, str) and os.path.exists(video):
             os.rename(video, temp_path)
         else:
-            with open(temp_path, "wb") as f:
+            with open(temp_path, 'wb') as f:
                 f.write(video.read())
-        input_data = {"video_path": temp_path}
-        result = predict_fn(input_data, model_dict)
-        # Build HTML output
-        output_md = ""
-        for i, utt in enumerate(result["utterances"], 1):
-            output_md += f"<div class='utt-card'>"
-            output_md += f"<h4>Utterance {i}</h4>"
-            output_md += f"<p><strong>Time:</strong> {utt['start_time']}s – {utt['end_time']}s</p>"
-            output_md += f"<p><strong>Text:</strong> {utt['text']}</p>"
-            # Emotions
-            output_md += "<div class='tag-group'><span class='tag-header'>Emotions:</span>"
-            for emo in utt['emotions']:
-                output_md += f"<span class='tag emo'>{emo['label'].capitalize()} ({emo['confidence']:.2f})</span>"
-            output_md += "</div>"
-            # Sentiments
-            output_md += "<div class='tag-group'><span class='tag-header'>Sentiments:</span>"
-            for sent in utt['sentiments']:
-                output_md += f"<span class='tag senti'>{sent['label'].capitalize()} ({sent['confidence']:.2f})</span>"
-            output_md += "</div>"
-            output_md += "</div>"
-        return output_md
+        # run inference
+        result = predict_fn({'video_path': temp_path}, model_dict)
+        utterances = result.get('utterances', [])
+        # overall analysis
+        overall = compute_overall(utterances) or {}
+        overall_html = f"""
+        <div class='overall-card'>
+            <div>
+                <strong>Primary Emotion</strong><br>
+                <span class='emoji'>{overall.get('emotion_emoji')}</span> {overall.get('emotion_label', '')}<br>
+                <small>{overall.get('emotion_score', 0)*100:.1f}%</small>
+            </div>
+            <div>
+                <strong>Primary Sentiment</strong><br>
+                <span class='emoji'>{overall.get('sentiment_emoji')}</span> {overall.get('sentiment_label', '')}<br>
+                <small>{overall.get('sentiment_score', 0)*100:.1f}%</small>
+            </div>
+        </div>
+        """
+        # utterance cards
+        utt_html = ''
+        for utt in utterances:
+            utt_html += f"""
+            <div class='utt-card'>
+                <div class='time'>{utt['start_time']:.1f}s - {utt['end_time']:.1f}s</div>
+                <div class='text'>{utt['text']}</div>
+                <div class='bar-group'><span>Emotions</span>"""
+            for emo in utt.get('emotions', []):
+                pct = emo['confidence'] * 100
+                utt_html += f"<div class='bar'><div class='fill emo' style='width:{pct:.0f}%'></div><span>{emo['label']}: {pct:.0f}%</span></div>"
+            utt_html += "</div><div class='bar-group'><span>Sentiments</span>"
+            for sent in utt.get('sentiments', []):
+                pct = sent['confidence'] * 100
+                utt_html += f"<div class='bar'><div class='fill senti' style='width:{pct:.0f}%'></div><span>{sent['label']}: {pct:.0f}%</span></div>"
+            utt_html += "</div></div>"
+        return overall_html + utt_html
     finally:
         if os.path.exists(temp_path):
             os.remove(temp_path)
 
-with gr.Blocks(css="""
-.utt-card {
-    background: #fff;
-    padding: 16px;
-    margin-bottom: 12px;
-    border-radius: 12px;
-    box-shadow: 0 2px 6px rgba(0,0,0,0.1);
-}
-.tag-group {
-    margin: 8px 0;
-    display: flex;
-    flex-wrap: wrap;
-    gap: 6px;
-    align-items: center;
-}
-.tag-header {
-    font-weight: 600;
-    margin-right: 6px;
-}
-.tag {
-    padding: 4px 8px;
-    border-radius: 8px;
-    font-size: 0.9em;
-}
-.tag.emo {
-    background: #fdecef;
-    color: #e74c3c;
-}
-.tag.senti {
-    background: #eafaf1;
-    color: #27ae60;
-}
-.gradio-container {
-    background-color: #f0f2f5;
-}
-#title {
-    font-size: 2rem;
-    font-weight: bold;
-    margin-bottom: 0.5em;
-}
-#description {
-    font-size: 1.1rem;
-    color: #555;
-}
-""") as demo:
-    gr.Markdown("# 🎬 Video Sentiment & Emotion Analyzer", elem_id="title")
-    gr.Markdown("Upload an .mp4 video to analyze emotions and sentiments per utterance.", elem_id="description")
+# Dark-mode CSS
+gr_css = """
+body { background: #121212; color: #e0e0e0; }
+.overall-card { display: flex; justify-content: space-around; background: #1f1f1f; padding: 16px; border-radius: 10px; margin-bottom: 16px; }
+.overall-card .emoji { font-size: 2rem; display: block; margin: 4px 0; }
+.utt-card { background: #1f1f1f; padding: 12px; margin-bottom: 12px; border-radius: 10px; }
+.time { font-size: 0.85rem; color: #888; margin-bottom: 4px; }
+.text { font-size: 0.95rem; margin-bottom: 8px; }
+.bar-group { margin-bottom: 8px; }
+.bar-group span { font-weight: 600; display: block; margin-bottom: 4px; }
+.bar { position: relative; background: #2a2a2a; height: 14px; border-radius: 7px; margin-bottom: 4px; overflow: hidden; }
+.fill { height: 100%; }
+.fill.emo { background: #e67e22; }
+.fill.senti { background: #27ae60; }
+.bar span { position: absolute; top: 0; left: 6px; font-size: 0.75rem; line-height: 14px; }
+"""
+
+# Build Gradio App
+with gr.Blocks(css=gr_css) as demo:
+    gr.Markdown('# 🎬 Video Sentiment & Emotion Analyzer')
     with gr.Row():
         with gr.Column(scale=1):
-            vid_input = gr.Video(label="Input Video")
-            run_btn = gr.Button("Analyze", variant="primary")
-        with gr.Column(scale=1):
-            output_html = gr.HTML(label="Results")
-
-    run_btn.click(fn=predict, inputs=vid_input, outputs=output_html)
+            vid_in = gr.Video(label='Upload a video')
+            btn = gr.Button('Analyze', variant='primary')
+        with gr.Column(scale=2):
+            out = gr.HTML()
+    btn.click(predict, inputs=vid_in, outputs=out)
 
-if __name__ == "__main__":
+if __name__ == '__main__':
     demo.launch()