fix: broadcast시 nparray 인코딩

- 이미지 업로드시 location name NONE
- Server 구동시 환경에 따라 CPU/GPU 설정 및 로그 출력

app/main.py

@@ -12,7 +12,7 @@
 from app.config import settings
 from app.connection_manager import ConnectionManager
 from app.history import async_save_history
-from model.detect import detect, track, estimate_distance, DetectionResult
+from model.detect import detect, estimate_distance, DetectionResult
 from model.video_recorder import VideoRecorder
 
 app = FastAPI()
@@ -36,7 +36,7 @@ async def detect_image(file: UploadFile = File(...)) -> ApiResponse:
         return ApiResponse.bad_request(str(err))
 
     result = detect(np.array(img))
-    await async_save_history(result)
+    await async_save_history(result, "NONE")
 
     return ApiResponse.ok()
 
@@ -65,7 +65,7 @@ async def websocket_publisher(websocket: WebSocket, location_name: str):
             if video_recorder.is_recording:
                 video_recorder.record_frame(result.plot_image)
 
-            await manager.broadcast(location_name, result.plot_image.tobytes())
+            await manager.broadcast(location_name, result.get_encoded_nparr().tobytes())
     except WebSocketDisconnect:
         manager.disconnect(location_name)
         print("Publisher disconnected")

model/detect.py

@@ -1,7 +1,9 @@
+import logging
 from typing import List, Dict, Tuple
 
 import cv2
 import numpy as np
+import torch
 from PIL import Image as img
 from numpy import ndarray
 from ultralytics import YOLO
@@ -11,6 +13,10 @@
 
 model = YOLO(settings.yolo_weight_path)
 
+device = 'cuda' if torch.cuda.is_available() else 'cpu'
+model.to(device)
+print(f"Model run on the {device}")
+
 tracked_objects: Dict[int, TrackedObject] = {}
 
 

model/schema.py

@@ -1,7 +1,9 @@
 from typing import List, Optional
 
+import cv2
 import numpy as np
 from PIL import Image as img
+from numpy import ndarray
 
 
 class TrackedObject:
@@ -37,3 +39,7 @@ def __init__(self, plot_image: np.ndarray, detections: List[Detection]):
 
     def get_image(self) -> img:
         return img.fromarray(self.plot_image[..., ::-1])
+
+    def get_encoded_nparr(self) -> ndarray:
+        _, encoded_nparr = cv2.imencode('.jpg', self.plot_image)
+        return encoded_nparr