dgenerate-ultralytics-headless 8.3.225__tar.gz → 8.3.226__tar.gz

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dgenerate-ultralytics-headless
-Version: 8.3.225
+Version: 8.3.226
 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>
@@ -160,7 +160,7 @@ Request an Enterprise License for commercial use at [Ultralytics Licensing](http
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
   <a href="https://twitter.com/ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-twitter.png" width="2%" alt="Ultralytics Twitter"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
-  <a href="https://youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="2%" alt="Ultralytics YouTube"></a>
+  <a href="https://www.youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="2%" alt="Ultralytics YouTube"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
   <a href="https://www.tiktok.com/@ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-tiktok.png" width="2%" alt="Ultralytics TikTok"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
@@ -396,7 +396,7 @@ For bug reports and feature requests related to Ultralytics software, please vis
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">
   <a href="https://twitter.com/ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-twitter.png" width="3%" alt="Ultralytics Twitter"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">
-  <a href="https://youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="3%" alt="Ultralytics YouTube"></a>
+  <a href="https://www.youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="3%" alt="Ultralytics YouTube"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">
   <a href="https://www.tiktok.com/@ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-tiktok.png" width="3%" alt="Ultralytics TikTok"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/README.md

@@ -70,7 +70,7 @@ Request an Enterprise License for commercial use at [Ultralytics Licensing](http
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
   <a href="https://twitter.com/ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-twitter.png" width="2%" alt="Ultralytics Twitter"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
-  <a href="https://youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="2%" alt="Ultralytics YouTube"></a>
+  <a href="https://www.youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="2%" alt="Ultralytics YouTube"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
   <a href="https://www.tiktok.com/@ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-tiktok.png" width="2%" alt="Ultralytics TikTok"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
@@ -306,7 +306,7 @@ For bug reports and feature requests related to Ultralytics software, please vis
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">
   <a href="https://twitter.com/ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-twitter.png" width="3%" alt="Ultralytics Twitter"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">
-  <a href="https://youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="3%" alt="Ultralytics YouTube"></a>
+  <a href="https://www.youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="3%" alt="Ultralytics YouTube"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">
   <a href="https://www.tiktok.com/@ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-tiktok.png" width="3%" alt="Ultralytics TikTok"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/dgenerate_ultralytics_headless.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dgenerate-ultralytics-headless
-Version: 8.3.225
+Version: 8.3.226
 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>
@@ -160,7 +160,7 @@ Request an Enterprise License for commercial use at [Ultralytics Licensing](http
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
   <a href="https://twitter.com/ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-twitter.png" width="2%" alt="Ultralytics Twitter"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
-  <a href="https://youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="2%" alt="Ultralytics YouTube"></a>
+  <a href="https://www.youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="2%" alt="Ultralytics YouTube"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
   <a href="https://www.tiktok.com/@ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-tiktok.png" width="2%" alt="Ultralytics TikTok"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="2%" alt="space">
@@ -396,7 +396,7 @@ For bug reports and feature requests related to Ultralytics software, please vis
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">
   <a href="https://twitter.com/ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-twitter.png" width="3%" alt="Ultralytics Twitter"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">
-  <a href="https://youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="3%" alt="Ultralytics YouTube"></a>
+  <a href="https://www.youtube.com/ultralytics?sub_confirmation=1"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-youtube.png" width="3%" alt="Ultralytics YouTube"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">
   <a href="https://www.tiktok.com/@ultralytics"><img src="https://github.com/ultralytics/assets/raw/main/social/logo-social-tiktok.png" width="3%" alt="Ultralytics TikTok"></a>
   <img src="https://github.com/ultralytics/assets/raw/main/social/logo-transparent.png" width="3%" alt="space">

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/tests/test_cuda.py

@@ -118,6 +118,7 @@ def test_train():
         results = YOLO(MODEL).train(
             data="coco8.yaml", imgsz=64, epochs=1, device=device, batch=15
         )  # requires imgsz>=64
+        results = YOLO(MODEL).train(data="coco8.yaml", imgsz=64, epochs=1, device=device, batch=15, val=False)
         visible = eval(os.environ["CUDA_VISIBLE_DEVICES"])
         assert visible == device, f"Passed GPUs '{device}', but used GPUs '{visible}'"
         assert (

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/tests/test_python.py

@@ -385,7 +385,46 @@ def test_cfg_init():
         check_dict_alignment({"a": 1}, {"b": 2})
     copy_default_cfg()
     (Path.cwd() / DEFAULT_CFG_PATH.name.replace(".yaml", "_copy.yaml")).unlink(missing_ok=False)
-    [smart_value(x) for x in {"none", "true", "false"}]
+
+    # Test smart_value() with comprehensive cases
+    # Test None conversion
+    assert smart_value("none") is None
+    assert smart_value("None") is None
+    assert smart_value("NONE") is None
+
+    # Test boolean conversion
+    assert smart_value("true") is True
+    assert smart_value("True") is True
+    assert smart_value("TRUE") is True
+    assert smart_value("false") is False
+    assert smart_value("False") is False
+    assert smart_value("FALSE") is False
+
+    # Test numeric conversion (ast.literal_eval)
+    assert smart_value("42") == 42
+    assert smart_value("-42") == -42
+    assert smart_value("3.14") == 3.14
+    assert smart_value("-3.14") == -3.14
+    assert smart_value("1e-3") == 0.001
+
+    # Test list/tuple conversion (ast.literal_eval)
+    assert smart_value("[1, 2, 3]") == [1, 2, 3]
+    assert smart_value("(1, 2, 3)") == (1, 2, 3)
+    assert smart_value("[640, 640]") == [640, 640]
+
+    # Test dict conversion (ast.literal_eval)
+    assert smart_value("{'a': 1, 'b': 2}") == {"a": 1, "b": 2}
+
+    # Test string fallback (when ast.literal_eval fails)
+    assert smart_value("some_string") == "some_string"
+    assert smart_value("path/to/file") == "path/to/file"
+    assert smart_value("hello world") == "hello world"
+
+    # Test that code injection is prevented (ast.literal_eval safety)
+    # These should return strings, not execute code
+    assert smart_value("__import__('os').system('ls')") == "__import__('os').system('ls')"
+    assert smart_value("eval('1+1')") == "eval('1+1')"
+    assert smart_value("exec('x=1')") == "exec('x=1')"
 
 
 def test_utils_init():
@@ -721,7 +760,7 @@ def test_grayscale(task: str, model: str, data: str, tmp_path) -> None:
     grayscale_data = tmp_path / f"{Path(data).stem}-grayscale.yaml"
     data = check_det_dataset(data)
     data["channels"] = 1  # add additional channels key for grayscale
-    YAML.save(grayscale_data, data)
+    YAML.save(data=data, file=grayscale_data)
     # remove npy files in train/val splits if exists, might be created by previous tests
     for split in {"train", "val"}:
         for npy_file in (Path(data["path"]) / data[split]).glob("*.npy"):

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/__init__.py

@@ -1,6 +1,6 @@
 # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
 
-__version__ = "8.3.225"
+__version__ = "8.3.226"
 
 import importlib
 import os

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/cfg/__init__.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import ast
 import shutil
 import subprocess
 import sys
@@ -459,7 +460,9 @@ def _handle_deprecation(custom: dict) -> dict:
     return custom
 
 
-def check_dict_alignment(base: dict, custom: dict, e: Exception | None = None) -> None:
+def check_dict_alignment(
+    base: dict, custom: dict, e: Exception | None = None, allowed_custom_keys: set | None = None
+) -> None:
     """Check alignment between custom and base configuration dictionaries, handling deprecated keys and providing error
     messages for mismatched keys.
 
@@ -467,6 +470,7 @@ def check_dict_alignment(base: dict, custom: dict, e: Exception | None = None) -
         base (dict): The base configuration dictionary containing valid keys.
         custom (dict): The custom configuration dictionary to be checked for alignment.
         e (Exception | None): Optional error instance passed by the calling function.
+        allowed_custom_keys (set | None): Optional set of additional keys that are allowed in the custom dictionary.
 
     Raises:
         SystemExit: If mismatched keys are found between the custom and base dictionaries.
@@ -486,7 +490,10 @@ def check_dict_alignment(base: dict, custom: dict, e: Exception | None = None) -
     """
     custom = _handle_deprecation(custom)
     base_keys, custom_keys = (frozenset(x.keys()) for x in (base, custom))
-    if mismatched := [k for k in custom_keys if k not in base_keys]:
+    # Allow 'augmentations' as a valid custom parameter for custom Albumentations transforms
+    if allowed_custom_keys is None:
+        allowed_custom_keys = {"augmentations"}
+    if mismatched := [k for k in custom_keys if k not in base_keys and k not in allowed_custom_keys]:
         from difflib import get_close_matches
 
         string = ""
@@ -797,7 +804,7 @@ def smart_value(v: str) -> Any:
 
     Notes:
         - The function uses a case-insensitive comparison for boolean and None values.
-        - For other types, it attempts to use Python's eval() function, which can be unsafe if used on untrusted input.
+        - For other types, it attempts to use Python's ast.literal_eval() function for safe evaluation.
         - If no conversion is possible, the original string is returned.
     """
     v_lower = v.lower()
@@ -809,7 +816,7 @@ def smart_value(v: str) -> Any:
         return False
     else:
         try:
-            return eval(v)
+            return ast.literal_eval(v)
         except Exception:
             return v
 

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/data/augment.py

@@ -1845,7 +1845,7 @@ class Albumentations:
         - Spatial transforms are handled differently and require special processing for bounding boxes.
     """
 
-    def __init__(self, p: float = 1.0) -> None:
+    def __init__(self, p: float = 1.0, transforms: list | None = None) -> None:
         """Initialize the Albumentations transform object for YOLO bbox formatted parameters.
 
         This class applies various image augmentations using the Albumentations library, including Blur, Median Blur,
@@ -1854,6 +1854,7 @@ class Albumentations:
 
         Args:
             p (float): Probability of applying the augmentations. Must be between 0 and 1.
+            transforms (list, optional): List of custom Albumentations transforms. If None, uses default transforms.
 
         Attributes:
             p (float): Probability of applying the augmentations.
@@ -1870,6 +1871,11 @@ class Albumentations:
             >>> augmented_image = augmented["image"]
             >>> augmented_bboxes = augmented["bboxes"]
 
+            >>> # Custom transforms example
+            >>> import albumentations as A
+            >>> custom_transforms = [A.Blur(p=0.01), A.CLAHE(p=0.01)]
+            >>> transform = Albumentations(p=1.0, transforms=custom_transforms)
+
         Notes:
             - Requires Albumentations version 1.0.3 or higher.
             - Spatial transforms are handled differently to ensure bbox compatibility.
@@ -1931,16 +1937,20 @@ class Albumentations:
                 "XYMasking",
             }  # from https://albumentations.ai/docs/getting_started/transforms_and_targets/#spatial-level-transforms
 
-            # Transforms
-            T = [
-                A.Blur(p=0.01),
-                A.MedianBlur(p=0.01),
-                A.ToGray(p=0.01),
-                A.CLAHE(p=0.01),
-                A.RandomBrightnessContrast(p=0.0),
-                A.RandomGamma(p=0.0),
-                A.ImageCompression(quality_range=(75, 100), p=0.0),
-            ]
+            # Transforms, use custom transforms if provided, otherwise use defaults
+            T = (
+                [
+                    A.Blur(p=0.01),
+                    A.MedianBlur(p=0.01),
+                    A.ToGray(p=0.01),
+                    A.CLAHE(p=0.01),
+                    A.RandomBrightnessContrast(p=0.0),
+                    A.RandomGamma(p=0.0),
+                    A.ImageCompression(quality_range=(75, 100), p=0.0),
+                ]
+                if transforms is None
+                else transforms
+            )
 
             # Compose transforms
             self.contains_spatial = any(transform.__class__.__name__ in spatial_transforms for transform in T)
@@ -2482,6 +2492,12 @@ def v8_transforms(dataset, imgsz: int, hyp: IterableSimpleNamespace, stretch: bo
         >>> hyp = IterableSimpleNamespace(mosaic=1.0, copy_paste=0.5, degrees=10.0, translate=0.2, scale=0.9)
         >>> transforms = v8_transforms(dataset, imgsz=640, hyp=hyp)
         >>> augmented_data = transforms(dataset[0])
+
+        >>> # With custom albumentations
+        >>> import albumentations as A
+        >>> augmentations = [A.Blur(p=0.01), A.CLAHE(p=0.01)]
+        >>> hyp.augmentations = augmentations
+        >>> transforms = v8_transforms(dataset, imgsz=640, hyp=hyp)
     """
     mosaic = Mosaic(dataset, imgsz=imgsz, p=hyp.mosaic)
     affine = RandomPerspective(
@@ -2519,7 +2535,7 @@ def v8_transforms(dataset, imgsz: int, hyp: IterableSimpleNamespace, stretch: bo
             pre_transform,
             MixUp(dataset, pre_transform=pre_transform, p=hyp.mixup),
             CutMix(dataset, pre_transform=pre_transform, p=hyp.cutmix),
-            Albumentations(p=1.0),
+            Albumentations(p=1.0, transforms=getattr(hyp, "augmentations", None)),
             RandomHSV(hgain=hyp.hsv_h, sgain=hyp.hsv_s, vgain=hyp.hsv_v),
             RandomFlip(direction="vertical", p=hyp.flipud, flip_idx=flip_idx),
             RandomFlip(direction="horizontal", p=hyp.fliplr, flip_idx=flip_idx),

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/data/loaders.py

@@ -123,7 +123,7 @@ class LoadStreams:
             if urllib.parse.urlparse(s).hostname in {"www.youtube.com", "youtube.com", "youtu.be"}:  # YouTube video
                 # YouTube format i.e. 'https://www.youtube.com/watch?v=Jsn8D3aC840' or 'https://youtu.be/Jsn8D3aC840'
                 s = get_best_youtube_url(s)
-            s = eval(s) if s.isnumeric() else s  # i.e. s = '0' local webcam
+            s = int(s) if s.isnumeric() else s  # i.e. s = '0' local webcam
             if s == 0 and (IS_COLAB or IS_KAGGLE):
                 raise NotImplementedError(
                     "'source=0' webcam not supported in Colab and Kaggle notebooks. "

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/engine/exporter.py

@@ -90,7 +90,6 @@ from ultralytics.utils import (
     MACOS,
     MACOS_VERSION,
     RKNN_CHIPS,
-    ROOT,
     SETTINGS,
     TORCH_VERSION,
     WINDOWS,
@@ -101,13 +100,11 @@ from ultralytics.utils import (
 )
 from ultralytics.utils.checks import (
     check_imgsz,
-    check_is_path_safe,
     check_requirements,
     check_version,
     is_intel,
     is_sudo_available,
 )
-from ultralytics.utils.downloads import get_github_assets, safe_download
 from ultralytics.utils.export import (
     keras2pb,
     onnx2engine,
@@ -165,12 +162,12 @@ def export_formats():
 
 def best_onnx_opset(onnx, cuda=False) -> int:
     """Return max ONNX opset for this torch version with ONNX fallback."""
-    version = ".".join(TORCH_VERSION.split(".")[:2])
     if TORCH_2_4:  # _constants.ONNX_MAX_OPSET first defined in torch 1.13
         opset = torch.onnx.utils._constants.ONNX_MAX_OPSET - 1  # use second-latest version for safety
         if cuda:
             opset -= 2  # fix CUDA ONNXRuntime NMS squeeze op errors
     else:
+        version = ".".join(TORCH_VERSION.split(".")[:2])
         opset = {
             "1.8": 12,
             "1.9": 12,
@@ -532,7 +529,7 @@ class Exporter:
             f[0] = self.export_torchscript()
         if engine:  # TensorRT required before ONNX
             f[1] = self.export_engine(dla=dla)
-        if onnx or ncnn:  # ONNX
+        if onnx:  # ONNX
             f[2] = self.export_onnx()
         if xml:  # OpenVINO
             f[3] = self.export_openvino()
@@ -822,65 +819,31 @@ class Exporter:
     def export_ncnn(self, prefix=colorstr("NCNN:")):
         """Export YOLO model to NCNN format using PNNX https://github.com/pnnx/pnnx."""
         check_requirements("ncnn", cmds="--no-deps")  # no deps to avoid installing opencv-python
+        check_requirements("pnnx")
         import ncnn
+        import pnnx
 
-        LOGGER.info(f"\n{prefix} starting export with NCNN {ncnn.__version__}...")
+        LOGGER.info(f"\n{prefix} starting export with NCNN {ncnn.__version__} and PNNX {pnnx.__version__}...")
         f = Path(str(self.file).replace(self.file.suffix, f"_ncnn_model{os.sep}"))
-        f_onnx = self.file.with_suffix(".onnx")
 
-        name = Path("pnnx.exe" if WINDOWS else "pnnx")  # PNNX filename
-        pnnx = name if name.is_file() else (ROOT / name)
-        if not pnnx.is_file():
-            LOGGER.warning(
-                f"{prefix} PNNX not found. Attempting to download binary file from "
-                "https://github.com/pnnx/pnnx/.\nNote PNNX Binary file must be placed in current working directory "
-                f"or in {ROOT}. See PNNX repo for full installation instructions."
-            )
-            system = "macos" if MACOS else "windows" if WINDOWS else "linux-aarch64" if ARM64 else "linux"
-            try:
-                release, assets = get_github_assets(repo="pnnx/pnnx")
-                asset = next(x for x in assets if f"{system}.zip" in x)
-                assert isinstance(asset, str), "Unable to retrieve PNNX repo assets"  # i.e. pnnx-20250930-macos.zip
-                LOGGER.info(f"{prefix} successfully found latest PNNX asset file {asset}")
-            except Exception as e:
-                release = "20250930"
-                asset = f"pnnx-{release}-{system}.zip"
-                LOGGER.warning(f"{prefix} PNNX GitHub assets not found: {e}, using default {asset}")
-            unzip_dir = safe_download(f"https://github.com/pnnx/pnnx/releases/download/{release}/{asset}", delete=True)
-            if check_is_path_safe(Path.cwd(), unzip_dir):  # avoid path traversal security vulnerability
-                shutil.move(src=unzip_dir / name, dst=pnnx)  # move binary to ROOT
-                pnnx.chmod(0o777)  # set read, write, and execute permissions for everyone
-                shutil.rmtree(unzip_dir)  # delete unzip dir
-
-        ncnn_args = [
-            f"ncnnparam={f / 'model.ncnn.param'}",
-            f"ncnnbin={f / 'model.ncnn.bin'}",
-            f"ncnnpy={f / 'model_ncnn.py'}",
-        ]
-
-        pnnx_args = [
-            f"pnnxparam={f / 'model.pnnx.param'}",
-            f"pnnxbin={f / 'model.pnnx.bin'}",
-            f"pnnxpy={f / 'model_pnnx.py'}",
-            f"pnnxonnx={f / 'model.pnnx.onnx'}",
-        ]
-
-        cmd = [
-            str(pnnx),
-            str(f_onnx),
-            *ncnn_args,
-            *pnnx_args,
-            f"fp16={int(self.args.half)}",
-            f"device={self.device.type}",
-            f'inputshape="{[self.args.batch, 3, *self.imgsz]}"',
-        ]
+        ncnn_args = dict(
+            ncnnparam=(f / "model.ncnn.param").as_posix(),
+            ncnnbin=(f / "model.ncnn.bin").as_posix(),
+            ncnnpy=(f / "model_ncnn.py").as_posix(),
+        )
+
+        pnnx_args = dict(
+            ptpath=(f / "model.pt").as_posix(),
+            pnnxparam=(f / "model.pnnx.param").as_posix(),
+            pnnxbin=(f / "model.pnnx.bin").as_posix(),
+            pnnxpy=(f / "model_pnnx.py").as_posix(),
+            pnnxonnx=(f / "model.pnnx.onnx").as_posix(),
+        )
+
         f.mkdir(exist_ok=True)  # make ncnn_model directory
-        LOGGER.info(f"{prefix} running '{' '.join(cmd)}'")
-        subprocess.run(cmd, check=True)
+        pnnx.export(self.model, inputs=self.im, **ncnn_args, **pnnx_args, fp16=self.args.half, device=self.device.type)
 
-        # Remove debug files
-        pnnx_files = [x.rsplit("=", 1)[-1] for x in pnnx_args]
-        for f_debug in ("debug.bin", "debug.param", "debug2.bin", "debug2.param", *pnnx_files):
+        for f_debug in ("debug.bin", "debug.param", "debug2.bin", "debug2.param", *pnnx_args.values()):
             Path(f_debug).unlink(missing_ok=True)
 
         YAML.save(f / "metadata.yaml", self.metadata)  # add metadata.yaml

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/engine/model.py

@@ -740,6 +740,7 @@ class Model(torch.nn.Module):
                 - optimizer (str): Optimizer to use for training.
                 - lr0 (float): Initial learning rate.
                 - patience (int): Epochs to wait for no observable improvement for early stopping of training.
+                - augmentations (list[Callable]): List of augmentation functions to apply during training.
 
         Returns:
             (dict | None): Training metrics if available and training is successful; otherwise, None.

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/engine/predictor.py

@@ -260,13 +260,12 @@ class BasePredictor:
             channels=getattr(self.model, "ch", 3),
         )
         self.source_type = self.dataset.source_type
-        long_sequence = (
+        if (
             self.source_type.stream
             or self.source_type.screenshot
             or len(self.dataset) > 1000  # many images
             or any(getattr(self.dataset, "video_flag", [False]))
-        )
-        if long_sequence:
+        ):  # long sequence
             import torchvision  # noqa (import here triggers torchvision NMS use in nms.py)
 
             if not getattr(self, "stream", True):  # videos

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/engine/trainer.py

@@ -138,7 +138,12 @@ class BaseTrainer:
         if RANK in {-1, 0}:
             self.wdir.mkdir(parents=True, exist_ok=True)  # make dir
             self.args.save_dir = str(self.save_dir)
-            YAML.save(self.save_dir / "args.yaml", vars(self.args))  # save run args
+            # Save run args, serializing augmentations as reprs for resume compatibility
+            args_dict = vars(self.args).copy()
+            if args_dict.get("augmentations") is not None:
+                # Serialize Albumentations transforms as their repr strings for checkpoint compatibility
+                args_dict["augmentations"] = [repr(t) for t in args_dict["augmentations"]]
+            YAML.save(self.save_dir / "args.yaml", args_dict)  # save run args
         self.last, self.best = self.wdir / "last.pt", self.wdir / "best.pt"  # checkpoint paths
         self.save_period = self.args.save_period
 
@@ -464,10 +469,10 @@ class BaseTrainer:
 
             self.run_callbacks("on_train_epoch_end")
             if RANK in {-1, 0}:
-                final_epoch = epoch + 1 >= self.epochs
                 self.ema.update_attr(self.model, include=["yaml", "nc", "args", "names", "stride", "class_weights"])
 
             # Validation
+            final_epoch = epoch + 1 >= self.epochs
             if self.args.val or final_epoch or self.stopper.possible_stop or self.stop:
                 self._clear_memory(threshold=0.5)  # prevent VRAM spike
                 self.metrics, self.fitness = self.validate()
@@ -626,7 +631,7 @@ class BaseTrainer:
         try:
             if self.args.task == "classify":
                 data = check_cls_dataset(self.args.data)
-            elif self.args.data.rsplit(".", 1)[-1] == "ndjson":
+            elif str(self.args.data).rsplit(".", 1)[-1] == "ndjson":
                 # Convert NDJSON to YOLO format
                 import asyncio
 
@@ -635,7 +640,7 @@ class BaseTrainer:
                 yaml_path = asyncio.run(convert_ndjson_to_yolo(self.args.data))
                 self.args.data = str(yaml_path)
                 data = check_det_dataset(self.args.data)
-            elif self.args.data.rsplit(".", 1)[-1] in {"yaml", "yml"} or self.args.task in {
+            elif str(self.args.data).rsplit(".", 1)[-1] in {"yaml", "yml"} or self.args.task in {
                 "detect",
                 "segment",
                 "pose",
@@ -755,9 +760,9 @@ class BaseTrainer:
         n = len(metrics) + 2  # number of cols
         t = time.time() - self.train_time_start
         self.csv.parent.mkdir(parents=True, exist_ok=True)  # ensure parent directory exists
-        s = "" if self.csv.exists() else (("%s," * n % tuple(["epoch", "time", *keys])).rstrip(",") + "\n")  # header
+        s = "" if self.csv.exists() else ("%s," * n % ("epoch", "time", *keys)).rstrip(",") + "\n"
         with open(self.csv, "a", encoding="utf-8") as f:
-            f.write(s + ("%.6g," * n % tuple([self.epoch + 1, t, *vals])).rstrip(",") + "\n")
+            f.write(s + ("%.6g," * n % (self.epoch + 1, t, *vals)).rstrip(",") + "\n")
 
     def plot_metrics(self):
         """Plot metrics from a CSV file."""
@@ -806,10 +811,21 @@ class BaseTrainer:
                     "batch",
                     "device",
                     "close_mosaic",
+                    "augmentations",
                 ):  # allow arg updates to reduce memory or update device on resume
                     if k in overrides:
                         setattr(self.args, k, overrides[k])
 
+                # Handle augmentations parameter for resume: check if user provided custom augmentations
+                if ckpt_args.get("augmentations") is not None:
+                    # Augmentations were saved in checkpoint as reprs but can't be restored automatically
+                    LOGGER.warning(
+                        "Custom Albumentations transforms were used in the original training run but are not "
+                        "being restored. To preserve custom augmentations when resuming, you need to pass the "
+                        "'augmentations' parameter again to get expected results. Example: \n"
+                        f"model.train(resume=True, augmentations={ckpt_args['augmentations']})"
+                    )
+
             except Exception as e:
                 raise FileNotFoundError(
                     "Resume checkpoint not found. Please pass a valid checkpoint to resume from, "

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/engine/tuner.py

@@ -295,8 +295,7 @@ class Tuner:
 
         # Try MongoDB first if available
         if self.mongodb:
-            results = self._get_mongodb_results(n)
-            if results:
+            if results := self._get_mongodb_results(n):
                 # MongoDB already sorted by fitness DESC, so results[0] is best
                 x = np.array([[r["fitness"]] + [r["hyperparameters"][k] for k in self.space.keys()] for r in results])
             elif self.collection.name in self.collection.database.list_collection_names():  # Tuner started elsewhere

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/models/fastsam/model.py

@@ -35,7 +35,7 @@ class FastSAM(Model):
         >>> results = model.predict("image.jpg", bboxes=[[100, 100, 200, 200]])
     """
 
-    def __init__(self, model: str = "FastSAM-x.pt"):
+    def __init__(self, model: str | Path = "FastSAM-x.pt"):
         """Initialize the FastSAM model with the specified pre-trained weights."""
         if str(model) == "FastSAM.pt":
             model = "FastSAM-x.pt"

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/models/yolo/detect/val.py

@@ -226,7 +226,7 @@ class DetectionValidator(BaseValidator):
             dist.gather_object(self.metrics.stats, gathered_stats, dst=0)
             merged_stats = {key: [] for key in self.metrics.stats.keys()}
             for stats_dict in gathered_stats:
-                for key in merged_stats.keys():
+                for key in merged_stats:
                     merged_stats[key].extend(stats_dict[key])
             gathered_jdict = [None] * dist.get_world_size()
             dist.gather_object(self.jdict, gathered_jdict, dst=0)

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/solutions/analytics.py

@@ -9,6 +9,7 @@ import cv2
 import numpy as np
 
 from ultralytics.solutions.solutions import BaseSolution, SolutionResults  # Import a parent class
+from ultralytics.utils import plt_settings
 
 
 class Analytics(BaseSolution):
@@ -46,6 +47,7 @@ class Analytics(BaseSolution):
         >>> cv2.imshow("Analytics", results.plot_im)
     """
 
+    @plt_settings()
     def __init__(self, **kwargs: Any) -> None:
         """Initialize Analytics class with various chart types for visual data representation."""
         super().__init__(**kwargs)

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/utils/__init__.py

@@ -722,14 +722,14 @@ def is_jetson(jetpack=None) -> bool:
     Returns:
         (bool): True if running on an NVIDIA Jetson device, False otherwise.
     """
-    if jetson := ("tegra" in DEVICE_MODEL):
-        if jetpack:
-            try:
-                content = open("/etc/nv_tegra_release").read()
-                version_map = {4: "R32", 5: "R35", 6: "R36"}  # JetPack to L4T major version mapping
-                return jetpack in version_map and version_map[jetpack] in content
-            except Exception:
-                return False
+    jetson = "tegra" in DEVICE_MODEL
+    if jetson and jetpack:
+        try:
+            content = open("/etc/nv_tegra_release").read()
+            version_map = {4: "R32", 5: "R35", 6: "R36"}  # JetPack to L4T major version mapping
+            return jetpack in version_map and version_map[jetpack] in content
+        except Exception:
+            return False
     return jetson
 
 

{dgenerate_ultralytics_headless-8.3.225 → dgenerate_ultralytics_headless-8.3.226}/ultralytics/utils/benchmarks.py

@@ -604,7 +604,7 @@ class ProfileModels:
         sess_options.intra_op_num_threads = 8  # Limit the number of threads
         sess = ort.InferenceSession(onnx_file, sess_options, providers=["CPUExecutionProvider"])
 
-        input_data_dict = dict()
+        input_data_dict = {}
         for input_tensor in sess.get_inputs():
             input_type = input_tensor.type
             if self.check_dynamic(input_tensor.shape):
@@ -632,7 +632,7 @@ class ProfileModels:
 
             input_data = np.random.rand(*input_shape).astype(input_dtype)
             input_name = input_tensor.name
-            input_data_dict.update({input_name: input_data})
+            input_data_dict[input_name] = input_data
 
         output_name = sess.get_outputs()[0].name