Simplify usage, remove YOLO.new() and YOLO.load()

tests/check_flops.py

@@ -4,7 +4,7 @@ from ultralytics.yolo.utils import ROOT
 if __name__ == "__main__":
     for m in list((ROOT / 'yolo/v8/models').rglob('*.yaml')):
         try:
-            YOLO.new(m.name, verbose=True)
+            YOLO(m.name, verbose=True)
         except Exception as e:
             print(f'ERROR for {m}: {e}')
 

tests/test_model.py

@@ -4,7 +4,7 @@ from ultralytics import YOLO
 
 
 def test_model_init():
-    model = YOLO.new("yolov8n.yaml")
+    model = YOLO("yolov8n.yaml")
     model.info()
     try:
         YOLO()
@@ -14,38 +14,38 @@ def test_model_init():
 
 
 def test_model_forward():
-    model = YOLO.new("yolov8n.yaml")
+    model = YOLO("yolov8n.yaml")
     img = torch.rand(512 * 512 * 3).view(1, 3, 512, 512)
     model.forward(img)
     model(img)
 
 
 def test_model_info():
-    model = YOLO.new("yolov8n.yaml")
+    model = YOLO("yolov8n.yaml")
     model.info()
     model = model.load("best.pt")
     model.info(verbose=True)
 
 
 def test_model_fuse():
-    model = YOLO.new("yolov8n.yaml")
+    model = YOLO("yolov8n.yaml")
     model.fuse()
     model.load("best.pt")
     model.fuse()
 
 
 def test_visualize_preds():
-    model = YOLO.load("best.pt")
+    model = YOLO("best.pt")
     model.predict(source="ultralytics/assets")
 
 
 def test_val():
-    model = YOLO.load("best.pt")
+    model = YOLO("best.pt")
     model.val(data="coco128.yaml", imgsz=32)
 
 
 def test_model_resume():
-    model = YOLO.new("yolov8n.yaml")
+    model = YOLO("yolov8n.yaml")
     model.train(epochs=1, imgsz=32, data="coco128.yaml")
     try:
         model.resume(task="detect")
@@ -54,7 +54,7 @@ def test_model_resume():
 
 
 def test_model_train_pretrained():
-    model = YOLO.load("best.pt")
+    model = YOLO("best.pt")
     model.train(data="coco128.yaml", epochs=1, imgsz=32)
     model = model.new("yolov8n.yaml")
     model.train(data="coco128.yaml", epochs=1, imgsz=32)
@@ -83,7 +83,7 @@ def test_exports():
 
     print(export_formats())
 
-    model = YOLO.new("yolov8n.yaml")
+    model = YOLO("yolov8n.yaml")
     model.export(format='torchscript')
     model.export(format='onnx')
     model.export(format='openvino')

ultralytics/yolo/engine/exporter.py

@@ -23,7 +23,7 @@ Requirements:
 
 Python:
     from ultralytics import YOLO
-    model = YOLO.new('yolov8n.yaml')
+    model = YOLO('yolov8n.yaml')
     results = model.export(format='onnx')
 
 CLI:

ultralytics/yolo/engine/model.py

@@ -1,4 +1,5 @@
 import torch
+from pathlib import Path
 
 from ultralytics import yolo  # noqa
 from ultralytics.nn.tasks import ClassificationModel, DetectionModel, SegmentationModel, attempt_load_weights
@@ -27,19 +28,15 @@ class YOLO:
 
     A python interface which emulates a model-like behaviour by wrapping trainers.
     """
-    __init_key = object()  # used to ensure proper initialization
 
-    def __init__(self, init_key=None, type="v8") -> None:
+    def __init__(self, model='yolov8n.yaml', type="v8") -> None:
         """
         Initializes the YOLO object.
 
         Args:
-            init_key (object): used to ensure proper initialization. Defaults to None.
+            model (str, Path): model to load or create
             type (str): Type/version of models to use. Defaults to "v8".
         """
-        if init_key != YOLO.__init_key:
-            raise SyntaxError(HELP_MSG)
-
         self.type = type
         self.ModelClass = None  # model class
         self.TrainerClass = None  # trainer class
@@ -53,8 +50,10 @@ class YOLO:
         self.overrides = {}  # overrides for trainer object
         self.init_disabled = False  # disable model initialization
 
-    @classmethod
-    def new(cls, cfg: str, verbose=True):
+        # Load or create new YOLO model
+        {'.pt': self._load, '.yaml': self._new}[Path(model).suffix](model)
+
+    def _new(self, cfg: str, verbose=True):
         """
         Initializes a new model and infers the task type from the model definitions.
 
@@ -64,34 +63,26 @@ class YOLO:
         """
         cfg = check_yaml(cfg)  # check YAML
         cfg_dict = yaml_load(cfg)  # model dict
-        obj = cls(init_key=cls.__init_key)
-        obj.task = guess_task_from_head(cfg_dict["head"][-1][-2])
-        obj.ModelClass, obj.TrainerClass, obj.ValidatorClass, obj.PredictorClass = obj._guess_ops_from_task(obj.task)
-        obj.model = obj.ModelClass(cfg_dict, verbose=verbose)  # initialize
-        obj.cfg = cfg
+        self.task = guess_task_from_head(cfg_dict["head"][-1][-2])
+        self.ModelClass, self.TrainerClass, self.ValidatorClass, self.PredictorClass = \
+            self._guess_ops_from_task(self.task)
+        self.model = self.ModelClass(cfg_dict, verbose=verbose)  # initialize
+        self.cfg = cfg
 
-        return obj
-
-    @classmethod
-    def load(cls, weights: str):
+    def _load(self, weights: str):
         """
         Initializes a new model and infers the task type from the model head
 
         Args:
             weights (str): model checkpoint to be loaded
         """
-        obj = cls(init_key=cls.__init_key)
-        obj.ckpt = torch.load(weights, map_location="cpu")
-        obj.task = obj.ckpt["train_args"]["task"]
-        obj.overrides = dict(obj.ckpt["train_args"])
-        obj.overrides["device"] = ''  # reset device
-        LOGGER.info("Device has been reset to ''")
-
-        obj.ModelClass, obj.TrainerClass, obj.ValidatorClass, obj.PredictorClass = obj._guess_ops_from_task(
-            task=obj.task)
-        obj.model = attempt_load_weights(weights)
-
-        return obj
+        self.ckpt = torch.load(weights, map_location="cpu")
+        self.task = self.ckpt["train_args"]["task"]
+        self.overrides = dict(self.ckpt["train_args"])
+        self.overrides["device"] = ''  # reset device
+        self.ModelClass, self.TrainerClass, self.ValidatorClass, self.PredictorClass = \
+            self._guess_ops_from_task(self.task)
+        self.model = attempt_load_weights(weights, fuse=False)
 
     def reset(self):
         """

ultralytics/yolo/utils/__init__.py

@@ -36,8 +36,8 @@ HELP_MSG = \
 
         from ultralytics import YOLO
 
-        model = YOLO.new('yolov8n.yaml')            # create a new model from scratch
-        model = YOLO.load('yolov8n.pt')             # load a pretrained model (recommended for best training results)
+        model = YOLO('yolov8n.yaml')                # build a new model from scratch
+        model = YOLO('yolov8n.pt')                  # load a pretrained model (recommended for best training results)
         results = model.train(data='coco128.yaml')  # train the model
         results = model.val()                       # evaluate model performance on the validation set
         results = model.predict(source='bus.jpg')   # predict on an image