Spaces:
Sleeping
Sleeping
Update model.py
Browse files
model.py
CHANGED
|
@@ -77,21 +77,15 @@ Prediction: **{text_label}**
|
|
| 77 |
|
| 78 |
Confidence: **{text_conf}%**
|
| 79 |
"""
|
| 80 |
-
|
| 81 |
except Exception as e:
|
| 82 |
text_result_display = f"Text error: {str(e)}"
|
| 83 |
|
| 84 |
# ================= IMAGE =================
|
| 85 |
-
image_path = None
|
| 86 |
if image is not None:
|
| 87 |
try:
|
| 88 |
-
|
| 89 |
-
# create image folder if not exists
|
| 90 |
-
|
| 91 |
results = image_pipeline(image)
|
| 92 |
|
| 93 |
image_result_display = "## ๐ผ Image Classification\n\n"
|
| 94 |
-
|
| 95 |
|
| 96 |
for r in results[:3]:
|
| 97 |
label = r["label"]
|
|
@@ -101,10 +95,6 @@ Confidence: **{text_conf}%**
|
|
| 101 |
image_label = results[0]["label"]
|
| 102 |
image_conf = round(results[0]["score"] * 100, 2)
|
| 103 |
|
| 104 |
-
# image preview
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
except Exception as e:
|
| 109 |
image_result_display = f"Image error: {str(e)}"
|
| 110 |
|
|
@@ -139,37 +129,103 @@ Confidence: **{audio_conf}%**
|
|
| 139 |
except Exception as e:
|
| 140 |
audio_result_display = f"Audio error: {str(e)}"
|
| 141 |
|
| 142 |
-
# ================= FUSION =================
|
| 143 |
-
|
| 144 |
-
|
| 145 |
-
|
| 146 |
-
|
| 147 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 148 |
|
| 149 |
processing_time = round(time.time() - start_time, 2)
|
| 150 |
|
|
|
|
|
|
|
| 151 |
fusion_summary = f"""
|
|
|
|
|
|
|
|
|
|
|
|
|
| 152 |
<h2>๐ Multimodal Intelligence Summary</h2>
|
| 153 |
|
| 154 |
-
{"<br>".join(
|
| 155 |
|
| 156 |
<hr>
|
| 157 |
|
| 158 |
-
<h3>Fusion Score</h3>
|
| 159 |
-
|
|
|
|
|
|
|
|
|
|
| 160 |
{round(fusion_score,2)}
|
| 161 |
</span>
|
|
|
|
| 162 |
<hr>
|
| 163 |
|
| 164 |
-
<h3>Interpretation</h3>
|
| 165 |
|
| 166 |
-
{
|
| 167 |
|
| 168 |
<br>
|
| 169 |
|
| 170 |
โฑ Processing Time: {processing_time} sec
|
|
|
|
|
|
|
| 171 |
"""
|
| 172 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 173 |
# ================= SAVE HISTORY =================
|
| 174 |
save_analysis({
|
| 175 |
"timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
|
|
|
|
| 77 |
|
| 78 |
Confidence: **{text_conf}%**
|
| 79 |
"""
|
|
|
|
| 80 |
except Exception as e:
|
| 81 |
text_result_display = f"Text error: {str(e)}"
|
| 82 |
|
| 83 |
# ================= IMAGE =================
|
|
|
|
| 84 |
if image is not None:
|
| 85 |
try:
|
|
|
|
|
|
|
|
|
|
| 86 |
results = image_pipeline(image)
|
| 87 |
|
| 88 |
image_result_display = "## ๐ผ Image Classification\n\n"
|
|
|
|
| 89 |
|
| 90 |
for r in results[:3]:
|
| 91 |
label = r["label"]
|
|
|
|
| 95 |
image_label = results[0]["label"]
|
| 96 |
image_conf = round(results[0]["score"] * 100, 2)
|
| 97 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
except Exception as e:
|
| 99 |
image_result_display = f"Image error: {str(e)}"
|
| 100 |
|
|
|
|
| 129 |
except Exception as e:
|
| 130 |
audio_result_display = f"Audio error: {str(e)}"
|
| 131 |
|
| 132 |
+
# ================= FUSION REASONING =================
|
| 133 |
+
|
| 134 |
+
reasoning_lines = []
|
| 135 |
+
|
| 136 |
+
if text_label:
|
| 137 |
+
reasoning_lines.append(
|
| 138 |
+
f"The textual input expresses a {text_label.lower()} emotional tone ({text_conf}% confidence)."
|
| 139 |
+
)
|
| 140 |
+
|
| 141 |
+
if image_label:
|
| 142 |
+
reasoning_lines.append(
|
| 143 |
+
f"The uploaded image is most likely '{image_label}' ({image_conf}% confidence)."
|
| 144 |
+
)
|
| 145 |
+
|
| 146 |
+
if transcription:
|
| 147 |
+
reasoning_lines.append(
|
| 148 |
+
"The audio transcription indicates spoken conversational content."
|
| 149 |
+
)
|
| 150 |
+
|
| 151 |
+
# ================= FUSION SCORE =================
|
| 152 |
+
|
| 153 |
+
fusion_score = 0
|
| 154 |
+
|
| 155 |
+
if text_label == "POSITIVE":
|
| 156 |
+
fusion_score += text_conf * 0.5
|
| 157 |
+
elif text_label == "NEGATIVE":
|
| 158 |
+
fusion_score -= text_conf * 0.5
|
| 159 |
+
|
| 160 |
+
if image_label:
|
| 161 |
+
fusion_score += image_conf * 0.3
|
| 162 |
+
|
| 163 |
+
if transcription:
|
| 164 |
+
fusion_score += 20
|
| 165 |
+
|
| 166 |
+
# ================= INTERPRETATION =================
|
| 167 |
+
|
| 168 |
+
if fusion_score > 60:
|
| 169 |
+
alignment_message = "Multimodal signals align toward a positive and confident contextual interpretation."
|
| 170 |
+
color = "#22c55e"
|
| 171 |
+
|
| 172 |
+
elif fusion_score < 0:
|
| 173 |
+
alignment_message = "Multimodal signals indicate a potentially negative contextual alignment."
|
| 174 |
+
color = "#ef4444"
|
| 175 |
+
|
| 176 |
+
else:
|
| 177 |
+
alignment_message = "Multimodal signals are contextually mixed or neutral."
|
| 178 |
+
color = "#f59e0b"
|
| 179 |
|
| 180 |
processing_time = round(time.time() - start_time, 2)
|
| 181 |
|
| 182 |
+
# ================= DISPLAY SUMMARY =================
|
| 183 |
+
|
| 184 |
fusion_summary = f"""
|
| 185 |
+
<div style="padding:20px;border-radius:16px;
|
| 186 |
+
background:linear-gradient(135deg,#0f172a,#1e293b);
|
| 187 |
+
border:1px solid #1f2a44;">
|
| 188 |
+
|
| 189 |
<h2>๐ Multimodal Intelligence Summary</h2>
|
| 190 |
|
| 191 |
+
{"<br>".join(reasoning_lines)}
|
| 192 |
|
| 193 |
<hr>
|
| 194 |
|
| 195 |
+
<h3>๐ Fusion Score</h3>
|
| 196 |
+
|
| 197 |
+
<span style="color:{color};
|
| 198 |
+
font-size:32px;
|
| 199 |
+
font-weight:700;">
|
| 200 |
{round(fusion_score,2)}
|
| 201 |
</span>
|
| 202 |
+
|
| 203 |
<hr>
|
| 204 |
|
| 205 |
+
<h3>๐ง Interpretation</h3>
|
| 206 |
|
| 207 |
+
{alignment_message}
|
| 208 |
|
| 209 |
<br>
|
| 210 |
|
| 211 |
โฑ Processing Time: {processing_time} sec
|
| 212 |
+
|
| 213 |
+
</div>
|
| 214 |
"""
|
| 215 |
|
| 216 |
+
# ================= SAVE HISTORY =================
|
| 217 |
+
|
| 218 |
+
save_analysis({
|
| 219 |
+
"timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
|
| 220 |
+
"text": text,
|
| 221 |
+
"image": image_label,
|
| 222 |
+
"audio": audio_label,
|
| 223 |
+
"transcription": transcription,
|
| 224 |
+
"fusion_score": round(fusion_score, 2)
|
| 225 |
+
})
|
| 226 |
+
|
| 227 |
+
return fusion_summary, text_result_display, image_result_display, audio_result_display
|
| 228 |
+
|
| 229 |
# ================= SAVE HISTORY =================
|
| 230 |
save_analysis({
|
| 231 |
"timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
|