refactor: output inference volume

app.py

@@ -76,7 +76,8 @@ def calculate_sphere_volume(width, height):
 @app.post("/inference_volume")
 async def inference_volume(file: UploadFile):
     """
-    Endpoint untuk deteksi tumor dengan volume, return image dengan anotasi
+    Endpoint sederhana untuk deteksi tumor dengan volume
+    Return: JSON dengan volume_mm3, class, dan image_bytes
     """
     if file.content_type not in ["image/jpeg", "image/png", "image/jpg"]:
         return JSONResponse(
@@ -91,11 +92,10 @@ async def inference_volume(file: UploadFile):
     results = model.predict(
         source=img, 
         conf=0.5,
-        iou=0.2
+        iou=0.45
     )
     
-    total_volume = 0
-    detection_count = 0
+    detections = []
     
     for r in results: 
         boxes = r.boxes
@@ -107,23 +107,23 @@ async def inference_volume(file: UploadFile):
             width = x2 - x1
             height = y2 - y1
             
-            volume = calculate_sphere_volume(width, height)
-            if volume is None:
-                volume = AVERAGE_TUMOR_VOLUME
-            
-            total_volume += volume
-            detection_count += 1
+            # Hitung volume
+            volume_mm3 = calculate_sphere_volume(width, height)
             
             label = int(box.cls[0].item())
             label_name = model.names[label]
-            confidence = box.conf[0].item()
+            
+            detections.append({
+                "class": label_name,
+                "volume_mm3": volume_mm3
+            })
             
             # Draw bounding box
             cv2.rectangle(img, (x1, y1), (x2, y2), (255, 0, 255), 2)
             
             # Label dengan volume
-            text = f"{label_name} {confidence:.2f}"
-            vol_text = f"Vol: {volume:.1f}mm3"
+            text = f"{label_name}"
+            vol_text = f"{volume_mm3} mm3"
             
             (text_width, text_height), baseline = cv2.getTextSize(
                 text, cv2.FONT_HERSHEY_SIMPLEX, 0.5, 2
@@ -132,7 +132,7 @@ async def inference_volume(file: UploadFile):
             cv2.rectangle(
                 img, 
                 (x1, y1 - text_height - baseline - 25), 
-                (x1 + max(text_width, 120), y1), 
+                (x1 + max(text_width, 100), y1), 
                 (255, 0, 255), 
                 -1
             )
@@ -146,15 +146,16 @@ async def inference_volume(file: UploadFile):
                 cv2.FONT_HERSHEY_SIMPLEX, 0.4, (255, 255, 255), 1
             )
     
-    summary_text = f"Total: {detection_count} tumor(s) | Vol: {total_volume:.1f}mm3"
-    cv2.rectangle(img, (10, 10), (400, 40), (0, 0, 0), -1)
-    cv2.putText(img, summary_text, (15, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 2)
+    # Encode image ke bytes
+    _, buffer = cv2.imencode('.jpg', img)
+    img_bytes = buffer.tobytes()
     
-    resp_img_bytes = cv2.imencode('.jpg', img)[1].tobytes()
-    resp_filename = f"volume_{file.filename}" if file.filename else "volume_image.jpg"
+    # Convert ke base64 untuk JSON
+    img_base64 = base64.b64encode(img_bytes).decode('utf-8')
+    
+    response = {
+        "detections": detections,
+        "image_bytes": img_base64
+    }
     
-    return StreamingResponse(
-        BytesIO(resp_img_bytes), 
-        media_type="image/jpeg", 
-        headers={"Content-Disposition": f"attachment; filename={resp_filename}"}
-    )
+    return JSONResponse(content=response)