Merge branch 'ai/feat/async' into 'ai/develop'

Feat: 오토레이블링, 학습 비동기 처리

See merge request s11-s-project/S11P21S002!230

ai/app/api/yolo/classfication.py

@@ -1,4 +1,5 @@
 from fastapi import APIRouter, HTTPException
+from fastapi.concurrency import run_in_threadpool
 from api.yolo.detection import run_predictions, get_random_color, split_data
 from schemas.predict_request import PredictRequest
 from schemas.train_request import TrainRequest, TrainDataInfo
@@ -25,7 +26,7 @@ async def classification_predict(request: PredictRequest):
     url_list = list(map(lambda x:x.image_url, request.image_list))
 
     # 추론
-    results = run_predictions(model, url_list, request, classes=[]) # classification은 classes를 무시함
+    results = await run_predictions(model, url_list, request, classes=[]) # classification은 classes를 무시함
 
     # 추론 결과 변환
     response = [process_prediction_result(result, image, request.label_map) for result, image in zip(results,request.image_list)]
@@ -104,7 +105,7 @@ async def classification_train(request: TrainRequest):
     download_data(train_data, test_data, dataset_root_path)
 
     # 학습
-    results = run_train(request, model,dataset_root_path)
+    results = await run_train(request, model,dataset_root_path)
 
     # best 모델 저장
     model_key = save_model(project_id=request.project_id, path=join_path(dataset_root_path, "result", "weights", "best.pt"))
@@ -136,7 +137,7 @@ def download_data(train_data:list[TrainDataInfo], test_data:list[TrainDataInfo],
         raise HTTPException(status_code=500, detail="exception in download_data(): " + str(e))
 
     
-def run_train(request, model, dataset_root_path):
+async def run_train(request, model, dataset_root_path):
     try:
         # 데이터 전송 콜백함수
         def send_data(trainer):
@@ -171,7 +172,7 @@ def run_train(request, model, dataset_root_path):
         model.add_callback("on_train_epoch_start", send_data)
 
         # 학습 실행
-        results = model.train(
+        results = await run_in_threadpool(model.train,
             data=dataset_root_path,
             name=join_path(dataset_root_path, "result"),
             epochs=request.epochs,

ai/app/api/yolo/detection.py

@@ -1,4 +1,5 @@
 from fastapi import APIRouter, HTTPException
+from fastapi.concurrency import run_in_threadpool
 from schemas.predict_request import PredictRequest
 from schemas.train_request import TrainRequest, TrainDataInfo
 from schemas.predict_response import PredictResponse, LabelData, Shape
@@ -29,7 +30,7 @@ async def detection_predict(request: PredictRequest):
     classes = get_classes(request.label_map, model.names)
 
     # 추론
-    results = run_predictions(model, url_list, request, classes)
+    results = await run_predictions(model, url_list, request, classes)
 
     # 추론 결과 변환
     response = [process_prediction_result(result, image, request.label_map) for result, image in zip(results,request.image_list)]
@@ -51,14 +52,16 @@ def get_classes(label_map:dict[str: int], model_names: dict[int, str]):
         raise HTTPException(status_code=500, detail="exception in get_classes(): " + str(e))
 
 # 추론 실행 함수
-def run_predictions(model, image, request, classes):
+async def run_predictions(model, image, request, classes):
     try:
-        return model.predict(
+        result = await run_in_threadpool(
+            model.predict, 
             source=image,
             iou=request.iou_threshold,
             conf=request.conf_threshold,
             classes=classes
         )
+        return result
     except Exception as e:
         raise HTTPException(status_code=500, detail="exception in run_predictions: " + str(e))
     
@@ -126,12 +129,13 @@ async def detection_train(request: TrainRequest):
 
     # 데이터 전처리: 데이터를 학습데이터와 검증데이터로 분류
     train_data, val_data = split_data(request.data, request.ratio)
+    
 
     # 데이터 전처리: 데이터 이미지 및 레이블 다운로드
     download_data(train_data, val_data, dataset_root_path, label_converter)
 
     # 학습
-    results = run_train(request, model,dataset_root_path)
+    results = await run_train(request, model,dataset_root_path)
 
     # best 모델 저장
     model_key = save_model(project_id=request.project_id, path=join_path(dataset_root_path, "result", "weights", "best.pt"))
@@ -157,6 +161,9 @@ def split_data(data:list[TrainDataInfo], ratio:float):
         random.shuffle(data)
         train_data = data[:train_size]
         val_data = data[train_size:]
+        
+        if not train_data or not val_data:
+            raise Exception("data size is too small")
         return train_data, val_data
     except Exception as e:
         raise HTTPException(status_code=500, detail="exception in split_data(): " + str(e))
@@ -171,7 +178,7 @@ def download_data(train_data:list[TrainDataInfo], val_data:list[TrainDataInfo],
     except Exception as e:
         raise HTTPException(status_code=500, detail="exception in download_data(): " + str(e))
     
-def run_train(request, model, dataset_root_path):
+async def run_train(request, model, dataset_root_path):
     try:
         # 데이터 전송 콜백함수
         def send_data(trainer):
@@ -206,7 +213,7 @@ def run_train(request, model, dataset_root_path):
         model.add_callback("on_train_epoch_start", send_data)
 
         # 학습 실행
-        results = model.train(
+        results =  await run_in_threadpool(model.train,
             data=join_path(dataset_root_path, "dataset.yaml"),
             name=join_path(dataset_root_path, "result"),
             epochs=request.epochs,

ai/app/api/yolo/segmentation.py

@@ -1,4 +1,5 @@
 from fastapi import APIRouter, HTTPException
+from fastapi.concurrency import run_in_threadpool
 from api.yolo.detection import get_classes, run_predictions, get_random_color, split_data, download_data
 from schemas.predict_request import PredictRequest
 from schemas.train_request import TrainRequest
@@ -27,7 +28,7 @@ async def segmentation_predict(request: PredictRequest):
     classes = get_classes(request.label_map, model.names)
 
     # 추론
-    results = run_predictions(model, url_list, request, classes)
+    results = await run_predictions(model, url_list, request, classes)
 
     # 추론 결과 변환
     response = [process_prediction_result(result, image, request.label_map) for result, image in zip(results,request.image_list)]
@@ -101,7 +102,7 @@ async def segmentation_train(request: TrainRequest):
     download_data(train_data, val_data, dataset_root_path, label_converter)
 
     # 학습
-    results = run_train(request, model,dataset_root_path)
+    results = await run_train(request, model,dataset_root_path)
 
     # best 모델 저장
     model_key = save_model(project_id=request.project_id, path=join_path(dataset_root_path, "result", "weights", "best.pt"))
@@ -121,7 +122,7 @@ async def segmentation_train(request: TrainRequest):
             
     return response
     
-def run_train(request, model, dataset_root_path):
+async def run_train(request, model, dataset_root_path):
     try:
         # 데이터 전송 콜백함수
         def send_data(trainer):
@@ -155,8 +156,9 @@ def run_train(request, model, dataset_root_path):
         # 콜백 등록
         model.add_callback("on_train_epoch_start", send_data)
 
+
         # 학습 실행
-        results = model.train(
+        results = await run_in_threadpool(model.train,
             data=join_path(dataset_root_path, "dataset.yaml"),
             name=join_path(dataset_root_path, "result"),
             epochs=request.epochs,