dgenerate-ultralytics-headless 8.3.240__tar.gz → 8.3.241__tar.gz

{dgenerate_ultralytics_headless-8.3.240 → dgenerate_ultralytics_headless-8.3.241}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dgenerate-ultralytics-headless
-Version: 8.3.240
+Version: 8.3.241
 Summary: Automatically built Ultralytics package with python-opencv-headless dependency instead of python-opencv
 Author-email: Glenn Jocher <glenn.jocher@ultralytics.com>, Jing Qiu <jing.qiu@ultralytics.com>
 Maintainer-email: Ultralytics <hello@ultralytics.com>
@@ -51,7 +51,7 @@ Requires-Dist: pytest; extra == "dev"
 Requires-Dist: pytest-cov; extra == "dev"
 Requires-Dist: coverage[toml]; extra == "dev"
 Requires-Dist: zensical>=0.0.9; python_version >= "3.10" and extra == "dev"
-Requires-Dist: mkdocs-ultralytics-plugin>=0.2.3; extra == "dev"
+Requires-Dist: mkdocs-ultralytics-plugin>=0.2.4; extra == "dev"
 Requires-Dist: minijinja>=2.0.0; extra == "dev"
 Provides-Extra: export
 Requires-Dist: numpy<2.0.0; extra == "export"

{dgenerate_ultralytics_headless-8.3.240 → dgenerate_ultralytics_headless-8.3.241}/dgenerate_ultralytics_headless.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dgenerate-ultralytics-headless
-Version: 8.3.240
+Version: 8.3.241
 Summary: Automatically built Ultralytics package with python-opencv-headless dependency instead of python-opencv
 Author-email: Glenn Jocher <glenn.jocher@ultralytics.com>, Jing Qiu <jing.qiu@ultralytics.com>
 Maintainer-email: Ultralytics <hello@ultralytics.com>
@@ -51,7 +51,7 @@ Requires-Dist: pytest; extra == "dev"
 Requires-Dist: pytest-cov; extra == "dev"
 Requires-Dist: coverage[toml]; extra == "dev"
 Requires-Dist: zensical>=0.0.9; python_version >= "3.10" and extra == "dev"
-Requires-Dist: mkdocs-ultralytics-plugin>=0.2.3; extra == "dev"
+Requires-Dist: mkdocs-ultralytics-plugin>=0.2.4; extra == "dev"
 Requires-Dist: minijinja>=2.0.0; extra == "dev"
 Provides-Extra: export
 Requires-Dist: numpy<2.0.0; extra == "export"

{dgenerate_ultralytics_headless-8.3.240 → dgenerate_ultralytics_headless-8.3.241}/dgenerate_ultralytics_headless.egg-info/requires.txt

@@ -19,7 +19,7 @@ ipython
 pytest
 pytest-cov
 coverage[toml]
-mkdocs-ultralytics-plugin>=0.2.3
+mkdocs-ultralytics-plugin>=0.2.4
 minijinja>=2.0.0
 
 [dev:python_version >= "3.10"]

{dgenerate_ultralytics_headless-8.3.240 → dgenerate_ultralytics_headless-8.3.241}/pyproject.toml

@@ -84,7 +84,7 @@ dev = [
     "pytest-cov",
     "coverage[toml]",
     "zensical>=0.0.9; python_version >= '3.10'",
-    "mkdocs-ultralytics-plugin>=0.2.3", # for meta descriptions and images, dates and authors
+    "mkdocs-ultralytics-plugin>=0.2.4", # for meta descriptions and images, dates and authors
     "minijinja>=2.0.0", # render docs macros without mkdocs-macros-plugin
 ]
 export = [

{dgenerate_ultralytics_headless-8.3.240 → dgenerate_ultralytics_headless-8.3.241}/ultralytics/__init__.py

@@ -1,6 +1,6 @@
 # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
 
-__version__ = "8.3.240"
+__version__ = "8.3.241"
 
 import importlib
 import os

{dgenerate_ultralytics_headless-8.3.240 → dgenerate_ultralytics_headless-8.3.241}/ultralytics/engine/exporter.py

@@ -1205,8 +1205,9 @@ class Exporter:
         # Setuptools bug: https://github.com/pypa/setuptools/issues/4483
         check_requirements("setuptools<71.0.0")  # Setuptools bug: https://github.com/pypa/setuptools/issues/4483
         check_requirements(("executorch==1.0.1", "flatbuffers"))
+        # Pin numpy to avoid coremltools errors with numpy>=2.4.0, must be separate
+        check_requirements("numpy<=2.3.5")
 
-        import torch
         from executorch.backends.xnnpack.partition.xnnpack_partitioner import XnnpackPartitioner
         from executorch.exir import to_edge_transform_and_lower
 

{dgenerate_ultralytics_headless-8.3.240 → dgenerate_ultralytics_headless-8.3.241}/ultralytics/nn/autobackend.py

@@ -245,15 +245,20 @@ class AutoBackend(nn.Module):
             check_requirements(("onnx", "onnxruntime-gpu" if cuda else "onnxruntime"))
             import onnxruntime
 
-            providers = ["CPUExecutionProvider"]
-            if cuda:
-                if "CUDAExecutionProvider" in onnxruntime.get_available_providers():
-                    providers.insert(0, ("CUDAExecutionProvider", {"device_id": device.index}))
-                else:  # Only log warning if CUDA was requested but unavailable
-                    LOGGER.warning("Failed to start ONNX Runtime with CUDA. Using CPU...")
-                    device = torch.device("cpu")
-                    cuda = False
-            LOGGER.info(f"Using ONNX Runtime {onnxruntime.__version__} {providers[0]}")
+            # Select execution provider: CUDA > CoreML (mps) > CPU
+            available = onnxruntime.get_available_providers()
+            if cuda and "CUDAExecutionProvider" in available:
+                providers = [("CUDAExecutionProvider", {"device_id": device.index}), "CPUExecutionProvider"]
+            elif device.type == "mps" and "CoreMLExecutionProvider" in available:
+                providers = ["CoreMLExecutionProvider", "CPUExecutionProvider"]
+            else:
+                providers = ["CPUExecutionProvider"]
+                if cuda:
+                    LOGGER.warning("CUDA requested but CUDAExecutionProvider not available. Using CPU...")
+                    device, cuda = torch.device("cpu"), False
+            LOGGER.info(
+                f"Using ONNX Runtime {onnxruntime.__version__} with {providers[0] if isinstance(providers[0], str) else providers[0][0]}"
+            )
             if onnx:
                 session = onnxruntime.InferenceSession(w, providers=providers)
             else:
@@ -271,7 +276,10 @@ class AutoBackend(nn.Module):
             metadata = session.get_modelmeta().custom_metadata_map
             dynamic = isinstance(session.get_outputs()[0].shape[0], str)
             fp16 = "float16" in session.get_inputs()[0].type
-            if not dynamic:
+
+            # Setup IO binding for optimized inference (CUDA only, not supported for CoreML)
+            use_io_binding = not dynamic and cuda
+            if use_io_binding:
                 io = session.io_binding()
                 bindings = []
                 for output in session.get_outputs():
@@ -371,32 +379,33 @@ class AutoBackend(nn.Module):
             is_trt10 = not hasattr(model, "num_bindings")
             num = range(model.num_io_tensors) if is_trt10 else range(model.num_bindings)
             for i in num:
+                # Get tensor info using TRT10+ or legacy API
                 if is_trt10:
                     name = model.get_tensor_name(i)
                     dtype = trt.nptype(model.get_tensor_dtype(name))
                     is_input = model.get_tensor_mode(name) == trt.TensorIOMode.INPUT
-                    if is_input:
-                        if -1 in tuple(model.get_tensor_shape(name)):
-                            dynamic = True
-                            context.set_input_shape(name, tuple(model.get_tensor_profile_shape(name, 0)[2]))
-                        if dtype == np.float16:
-                            fp16 = True
-                    else:
-                        output_names.append(name)
-                    shape = tuple(context.get_tensor_shape(name))
-                else:  # TensorRT < 10.0
+                    shape = tuple(model.get_tensor_shape(name))
+                    profile_shape = tuple(model.get_tensor_profile_shape(name, 0)[2]) if is_input else None
+                else:
                     name = model.get_binding_name(i)
                     dtype = trt.nptype(model.get_binding_dtype(i))
                     is_input = model.binding_is_input(i)
-                    if model.binding_is_input(i):
-                        if -1 in tuple(model.get_binding_shape(i)):  # dynamic
-                            dynamic = True
-                            context.set_binding_shape(i, tuple(model.get_profile_shape(0, i)[1]))
-                        if dtype == np.float16:
-                            fp16 = True
-                    else:
-                        output_names.append(name)
-                    shape = tuple(context.get_binding_shape(i))
+                    shape = tuple(model.get_binding_shape(i))
+                    profile_shape = tuple(model.get_profile_shape(0, i)[1]) if is_input else None
+
+                # Process input/output tensors
+                if is_input:
+                    if -1 in shape:
+                        dynamic = True
+                        if is_trt10:
+                            context.set_input_shape(name, profile_shape)
+                        else:
+                            context.set_binding_shape(i, profile_shape)
+                    if dtype == np.float16:
+                        fp16 = True
+                else:
+                    output_names.append(name)
+                shape = tuple(context.get_tensor_shape(name)) if is_trt10 else tuple(context.get_binding_shape(i))
                 im = torch.from_numpy(np.empty(shape, dtype=dtype)).to(device)
                 bindings[name] = Binding(name, dtype, shape, im, int(im.data_ptr()))
             binding_addrs = OrderedDict((n, d.ptr) for n, d in bindings.items())
@@ -418,8 +427,7 @@ class AutoBackend(nn.Module):
             LOGGER.info(f"Loading {w} for TensorFlow SavedModel inference...")
             import tensorflow as tf
 
-            keras = False  # assume TF1 saved_model
-            model = tf.keras.models.load_model(w) if keras else tf.saved_model.load(w)
+            model = tf.saved_model.load(w)
             metadata = Path(w) / "metadata.yaml"
 
         # TF GraphDef
@@ -699,10 +707,7 @@ class AutoBackend(nn.Module):
 
         # ONNX Runtime
         elif self.onnx or self.imx:
-            if self.dynamic:
-                im = im.cpu().numpy()  # torch to numpy
-                y = self.session.run(self.output_names, {self.session.get_inputs()[0].name: im})
-            else:
+            if self.use_io_binding:
                 if not self.cuda:
                     im = im.cpu()
                 self.io.bind_input(
@@ -715,6 +720,9 @@ class AutoBackend(nn.Module):
                 )
                 self.session.run_with_iobinding(self.io)
                 y = self.bindings
+            else:
+                im = im.cpu().numpy()  # torch to numpy
+                y = self.session.run(self.output_names, {self.session.get_inputs()[0].name: im})
             if self.imx:
                 if self.task == "detect":
                     # boxes, conf, cls
@@ -839,7 +847,7 @@ class AutoBackend(nn.Module):
         else:
             im = im.cpu().numpy()
             if self.saved_model:  # SavedModel
-                y = self.model(im, training=False) if self.keras else self.model.serving_default(im)
+                y = self.model.serving_default(im)
                 if not isinstance(y, list):
                     y = [y]
             elif self.pb:  # GraphDef
@@ -884,8 +892,6 @@ class AutoBackend(nn.Module):
                     y[1] = np.transpose(y[1], (0, 3, 1, 2))  # should be y = (1, 116, 8400), (1, 32, 160, 160)
             y = [x if isinstance(x, np.ndarray) else x.numpy() for x in y]
 
-        # for x in y:
-        #     print(type(x), len(x)) if isinstance(x, (list, tuple)) else print(type(x), x.shape)  # debug shapes
         if isinstance(y, (list, tuple)):
             if len(self.names) == 999 and (self.task == "segment" or len(y) == 2):  # segments and names not defined
                 nc = y[0].shape[1] - y[1].shape[1] - 4  # y = (1, 32, 160, 160), (1, 116, 8400)

{dgenerate_ultralytics_headless-8.3.240 → dgenerate_ultralytics_headless-8.3.241}/ultralytics/utils/__init__.py

@@ -65,6 +65,7 @@ RKNN_CHIPS = frozenset(
         "rv1103b",
         "rv1106b",
         "rk2118",
+        "rv1126b",
     }
 )  # Rockchip processors available for export
 HELP_MSG = """

{dgenerate_ultralytics_headless-8.3.240 → dgenerate_ultralytics_headless-8.3.241}/ultralytics/utils/plotting.py

@@ -203,6 +203,8 @@ class Annotator:
         if not input_is_pil:
             if im.shape[2] == 1:  # handle grayscale
                 im = cv2.cvtColor(im, cv2.COLOR_GRAY2BGR)
+            elif im.shape[2] == 2:  # handle 2-channel images
+                im = np.ascontiguousarray(np.dstack((im, np.zeros_like(im[..., :1]))))
             elif im.shape[2] > 3:  # multispectral
                 im = np.ascontiguousarray(im[..., :3])
         if self.pil:  # use PIL