dgenerate-ultralytics-headless 8.3.234__tar.gz → 8.3.235__tar.gz

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dgenerate-ultralytics-headless
-Version: 8.3.234
+Version: 8.3.235
 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>
@@ -57,7 +57,7 @@ Provides-Extra: export
 Requires-Dist: numpy<2.0.0; extra == "export"
 Requires-Dist: onnx>=1.12.0; platform_system != "Darwin" and extra == "export"
 Requires-Dist: onnx<1.18.0,>=1.12.0; platform_system == "Darwin" and extra == "export"
-Requires-Dist: coremltools>=8.0; (platform_system != "Windows" and python_version <= "3.13") and extra == "export"
+Requires-Dist: coremltools>=9.0; (platform_system != "Windows" and python_version <= "3.13") and extra == "export"
 Requires-Dist: scikit-learn>=1.3.2; (platform_system != "Windows" and python_version <= "3.13") and extra == "export"
 Requires-Dist: openvino>=2024.0.0; extra == "export"
 Requires-Dist: tensorflow<=2.19.0,>=2.0.0; extra == "export"

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/dgenerate_ultralytics_headless.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dgenerate-ultralytics-headless
-Version: 8.3.234
+Version: 8.3.235
 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>
@@ -57,7 +57,7 @@ Provides-Extra: export
 Requires-Dist: numpy<2.0.0; extra == "export"
 Requires-Dist: onnx>=1.12.0; platform_system != "Darwin" and extra == "export"
 Requires-Dist: onnx<1.18.0,>=1.12.0; platform_system == "Darwin" and extra == "export"
-Requires-Dist: coremltools>=8.0; (platform_system != "Windows" and python_version <= "3.13") and extra == "export"
+Requires-Dist: coremltools>=9.0; (platform_system != "Windows" and python_version <= "3.13") and extra == "export"
 Requires-Dist: scikit-learn>=1.3.2; (platform_system != "Windows" and python_version <= "3.13") and extra == "export"
 Requires-Dist: openvino>=2024.0.0; extra == "export"
 Requires-Dist: tensorflow<=2.19.0,>=2.0.0; extra == "export"

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/dgenerate_ultralytics_headless.egg-info/requires.txt

@@ -41,7 +41,7 @@ tensorstore>=0.1.63
 onnx>=1.12.0
 
 [export:platform_system != "Windows" and python_version <= "3.13"]
-coremltools>=8.0
+coremltools>=9.0
 scikit-learn>=1.3.2
 
 [export:platform_system == "Darwin"]

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/pyproject.toml

@@ -91,7 +91,7 @@ export = [
     "numpy<2.0.0", # TF 2.20 compatibility
     "onnx>=1.12.0; platform_system != 'Darwin'", # ONNX export
     "onnx>=1.12.0,<1.18.0; platform_system == 'Darwin'",  # TF inference hanging on MacOS
-    "coremltools>=8.0; platform_system != 'Windows' and python_version <= '3.13'", # CoreML supported on macOS and Linux
+    "coremltools>=9.0; platform_system != 'Windows' and python_version <= '3.13'", # CoreML supported on macOS and Linux
     "scikit-learn>=1.3.2; platform_system != 'Windows' and python_version <= '3.13'", # CoreML k-means quantization
     "openvino>=2024.0.0",  # OpenVINO export
     "tensorflow>=2.0.0,<=2.19.0", # TF bug https://github.com/ultralytics/ultralytics/issues/5161

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/tests/test_exports.py

@@ -248,11 +248,13 @@ def test_export_ncnn_matrix(task, half, batch):
     shutil.rmtree(file, ignore_errors=True)  # retry in case of potential lingering multi-threaded file usage errors
 
 
-@pytest.mark.skipif(True, reason="Test disabled as keras and tensorflow version conflicts with TFlite export.")
-@pytest.mark.skipif(not LINUX or MACOS, reason="Skipping test on Windows and Macos")
+@pytest.mark.skipif(not TORCH_2_9, reason="IMX export requires torch>=2.9.0")
+@pytest.mark.skipif(not checks.IS_PYTHON_MINIMUM_3_9, reason="Requires Python>=3.9")
+@pytest.mark.skipif(WINDOWS or MACOS, reason="Skipping test on Windows and Macos")
+@pytest.mark.skipif(ARM64, reason="IMX export is not supported on ARM64 architectures.")
 def test_export_imx():
     """Test YOLO export to IMX format."""
-    model = YOLO("yolov8n.pt")
+    model = YOLO(MODEL)
     file = model.export(format="imx", imgsz=32)
     YOLO(file)(SOURCE, imgsz=32)
 

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/ultralytics/__init__.py

@@ -1,6 +1,6 @@
 # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
 
-__version__ = "8.3.234"
+__version__ = "8.3.235"
 
 import importlib
 import os

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/ultralytics/engine/exporter.py

@@ -103,7 +103,7 @@ from ultralytics.utils import (
     get_default_args,
 )
 from ultralytics.utils.checks import (
-    IS_PYTHON_3_12,
+    IS_PYTHON_MINIMUM_3_9,
     check_imgsz,
     check_requirements,
     check_version,
@@ -365,11 +365,13 @@ class Exporter:
             if not self.args.int8:
                 LOGGER.warning("IMX export requires int8=True, setting int8=True.")
                 self.args.int8 = True
-            if not self.args.nms and model.task in {"detect", "pose"}:
+            if not self.args.nms and model.task in {"detect", "pose", "segment"}:
                 LOGGER.warning("IMX export requires nms=True, setting nms=True.")
                 self.args.nms = True
-            if model.task not in {"detect", "pose", "classify"}:
-                raise ValueError("IMX export only supported for detection, pose estimation, and classification models.")
+            if model.task not in {"detect", "pose", "classify", "segment"}:
+                raise ValueError(
+                    "IMX export only supported for detection, pose estimation, classification, and segmentation models."
+                )
         if not hasattr(model, "names"):
             model.names = default_class_names()
         model.names = check_class_names(model.names)
@@ -858,7 +860,9 @@ class Exporter:
     def export_coreml(self, prefix=colorstr("CoreML:")):
         """Export YOLO model to CoreML format."""
         mlmodel = self.args.format.lower() == "mlmodel"  # legacy *.mlmodel export format requested
-        check_requirements("coremltools>=8.0")
+        check_requirements(
+            ["coremltools>=9.0", "numpy>=1.14.5,<=2.3.5"]
+        )  # latest numpy 2.4.0rc1 breaks coremltools exports
         import coremltools as ct
 
         LOGGER.info(f"\n{prefix} starting export with coremltools {ct.__version__}...")
@@ -1086,7 +1090,7 @@ class Exporter:
         # TorchAO release compatibility table bug https://github.com/pytorch/ao/issues/2919
         # 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==1.0.0", "flatbuffers"))
+        check_requirements(("executorch==1.0.1", "flatbuffers"))
 
         import torch
         from executorch.backends.xnnpack.partition.xnnpack_partitioner import XnnpackPartitioner
@@ -1177,18 +1181,24 @@ class Exporter:
     def export_imx(self, prefix=colorstr("IMX:")):
         """Export YOLO model to IMX format."""
         assert LINUX, (
-            "export only supported on Linux. "
-            "See https://developer.aitrios.sony-semicon.com/en/raspberrypi-ai-camera/documentation/imx500-converter"
+            "Export only supported on Linux."
+            "See https://developer.aitrios.sony-semicon.com/en/docs/raspberry-pi-ai-camera/imx500-converter?version=3.17.3&progLang="
         )
-        assert not IS_PYTHON_3_12, "IMX export requires Python>=3.8;<3.12"
-        assert not TORCH_2_9, f"IMX export requires PyTorch<2.9. Current PyTorch version is {TORCH_VERSION}."
+        assert not ARM64, "IMX export is not supported on ARM64 architectures."
+        assert IS_PYTHON_MINIMUM_3_9, "IMX export is only supported on Python 3.9 or above."
+
         if getattr(self.model, "end2end", False):
             raise ValueError("IMX export is not supported for end2end models.")
         check_requirements(
-            ("model-compression-toolkit>=2.4.1", "sony-custom-layers>=0.3.0", "edge-mdt-tpc>=1.1.0", "pydantic<=2.11.7")
+            (
+                "model-compression-toolkit>=2.4.1",
+                "edge-mdt-cl<1.1.0",
+                "edge-mdt-tpc>=1.2.0",
+                "pydantic<=2.11.7",
+            )
         )
-        check_requirements("imx500-converter[pt]>=3.16.1")  # Separate requirements for imx500-converter
-        check_requirements("mct-quantizers>=1.6.0")  # Separate for compatibility with model-compression-toolkit
+
+        check_requirements("imx500-converter[pt]>=3.17.3")
 
         # Install Java>=17
         try:
@@ -1200,10 +1210,10 @@ class Exporter:
             cmd = None
             if IS_UBUNTU or IS_DEBIAN_TRIXIE:
                 LOGGER.info(f"\n{prefix} installing Java 21 for Ubuntu...")
-                cmd = (["sudo"] if is_sudo_available() else []) + ["apt", "install", "-y", "openjdk-21-jre"]
+                cmd = (["sudo"] if is_sudo_available() else []) + ["apt-get", "install", "-y", "openjdk-21-jre"]
             elif IS_RASPBERRYPI or IS_DEBIAN_BOOKWORM:
                 LOGGER.info(f"\n{prefix} installing Java 17 for Raspberry Pi or Debian ...")
-                cmd = (["sudo"] if is_sudo_available() else []) + ["apt", "install", "-y", "openjdk-17-jre"]
+                cmd = (["sudo"] if is_sudo_available() else []) + ["apt-get", "install", "-y", "openjdk-17-jre"]
 
             if cmd:
                 subprocess.run(cmd, check=True)

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/ultralytics/nn/autobackend.py

@@ -254,13 +254,11 @@ class AutoBackend(nn.Module):
             if onnx:
                 session = onnxruntime.InferenceSession(w, providers=providers)
             else:
-                check_requirements(
-                    ("model-compression-toolkit>=2.4.1", "sony-custom-layers[torch]>=0.3.0", "onnxruntime-extensions")
-                )
+                check_requirements(("model-compression-toolkit>=2.4.1", "edge-mdt-cl<1.1.0", "onnxruntime-extensions"))
                 w = next(Path(w).glob("*.onnx"))
                 LOGGER.info(f"Loading {w} for ONNX IMX inference...")
                 import mct_quantizers as mctq
-                from sony_custom_layers.pytorch.nms import nms_ort  # noqa
+                from edgemdt_cl.pytorch.nms import nms_ort  # noqa - register custom NMS ops
 
                 session_options = mctq.get_ort_session_options()
                 session_options.enable_mem_reuse = False  # fix the shape mismatch from onnxruntime
@@ -402,7 +400,9 @@ class AutoBackend(nn.Module):
 
         # CoreML
         elif coreml:
-            check_requirements("coremltools>=8.0")
+            check_requirements(
+                ["coremltools>=9.0", "numpy>=1.14.5,<=2.3.5"]
+            )  # latest numpy 2.4.0rc1 breaks coremltools exports
             LOGGER.info(f"Loading {w} for CoreML inference...")
             import coremltools as ct
 
@@ -575,7 +575,7 @@ class AutoBackend(nn.Module):
             LOGGER.info(f"Loading {w} for ExecuTorch inference...")
             # TorchAO release compatibility table bug https://github.com/pytorch/ao/issues/2919
             check_requirements("setuptools<71.0.0")  # Setuptools bug: https://github.com/pypa/setuptools/issues/4483
-            check_requirements(("executorch==1.0.0", "flatbuffers"))
+            check_requirements(("executorch==1.0.1", "flatbuffers"))
             from executorch.runtime import Runtime
 
             w = Path(w)
@@ -691,7 +691,12 @@ class AutoBackend(nn.Module):
                     y = np.concatenate([y[0], y[1][:, :, None], y[2][:, :, None]], axis=-1)
                 elif self.task == "pose":
                     # boxes, conf, kpts
-                    y = np.concatenate([y[0], y[1][:, :, None], y[2][:, :, None], y[3]], axis=-1)
+                    y = np.concatenate([y[0], y[1][:, :, None], y[2][:, :, None], y[3]], axis=-1, dtype=y[0].dtype)
+                elif self.task == "segment":
+                    y = (
+                        np.concatenate([y[0], y[1][:, :, None], y[2][:, :, None], y[3]], axis=-1, dtype=y[0].dtype),
+                        y[4],
+                    )
 
         # OpenVINO
         elif self.xml:

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/ultralytics/utils/callbacks/clearml.py

@@ -123,15 +123,9 @@ def on_train_end(trainer) -> None:
     """Log final model and training results on training completion."""
     if task := Task.current_task():
         # Log final results, confusion matrix and PR plots
-        files = [
-            "results.png",
-            "confusion_matrix.png",
-            "confusion_matrix_normalized.png",
-            *(f"{x}_curve.png" for x in ("F1", "PR", "P", "R")),
-        ]
-        files = [(trainer.save_dir / f) for f in files if (trainer.save_dir / f).exists()]  # filter existing files
-        for f in files:
-            _log_plot(title=f.stem, plot_path=f)
+        for f in [*trainer.plots.keys(), *trainer.validator.plots.keys()]:
+            if "batch" not in f.name:
+                _log_plot(title=f.stem, plot_path=f)
         # Report final metrics
         for k, v in trainer.validator.metrics.results_dict.items():
             task.get_logger().report_single_value(k, v)

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/ultralytics/utils/callbacks/neptune.py

@@ -106,15 +106,9 @@ def on_train_end(trainer) -> None:
     """Log final results, plots, and model weights at the end of training."""
     if run:
         # Log final results, CM matrix + PR plots
-        files = [
-            "results.png",
-            "confusion_matrix.png",
-            "confusion_matrix_normalized.png",
-            *(f"{x}_curve.png" for x in ("F1", "PR", "P", "R")),
-        ]
-        files = [(trainer.save_dir / f) for f in files if (trainer.save_dir / f).exists()]  # filter
-        for f in files:
-            _log_plot(title=f.stem, plot_path=f)
+        for f in [*trainer.plots.keys(), *trainer.validator.plots.keys()]:
+            if "batch" not in f.name:
+                _log_plot(title=f.stem, plot_path=f)
         # Log the final model
         run[f"weights/{trainer.args.name or trainer.args.task}/{trainer.best.name}"].upload(File(str(trainer.best)))
 

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/ultralytics/utils/checks.py

@@ -410,7 +410,7 @@ def check_requirements(requirements=ROOT.parent / "requirements.txt", exclude=()
         if use_uv:
             base = (
                 f"uv pip install --no-cache-dir {packages} {commands} "
-                f"--index-strategy=unsafe-best-match --break-system-packages --prerelease=allow"
+                f"--index-strategy=unsafe-best-match --break-system-packages"
             )
             try:
                 return subprocess.check_output(base, shell=True, stderr=subprocess.STDOUT, text=True)
@@ -947,8 +947,10 @@ check_torchvision()  # check torch-torchvision compatibility
 
 # Define constants
 IS_PYTHON_3_8 = PYTHON_VERSION.startswith("3.8")
+IS_PYTHON_3_9 = PYTHON_VERSION.startswith("3.9")
 IS_PYTHON_3_12 = PYTHON_VERSION.startswith("3.12")
 IS_PYTHON_3_13 = PYTHON_VERSION.startswith("3.13")
 
+IS_PYTHON_MINIMUM_3_9 = check_python("3.9", hard=False)
 IS_PYTHON_MINIMUM_3_10 = check_python("3.10", hard=False)
 IS_PYTHON_MINIMUM_3_12 = check_python("3.12", hard=False)

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/ultralytics/utils/export/imx.py

@@ -9,8 +9,9 @@ from pathlib import Path
 import numpy as np
 import torch
 
-from ultralytics.nn.modules import Detect, Pose
+from ultralytics.nn.modules import Detect, Pose, Segment
 from ultralytics.utils import LOGGER
+from ultralytics.utils.patches import onnx_export_patch
 from ultralytics.utils.tal import make_anchors
 from ultralytics.utils.torch_utils import copy_attr
 
@@ -28,6 +29,7 @@ MCT_CONFIG = {
             "n_layers": 257,
         },
         "classify": {"layer_names": [], "weights_memory": np.inf, "n_layers": 112},
+        "segment": {"layer_names": ["sub", "mul_2", "add_14", "cat_22"], "weights_memory": 2466604.8, "n_layers": 265},
     },
     "YOLOv8": {
         "detect": {"layer_names": ["sub", "mul", "add_6", "cat_17"], "weights_memory": 2550540.8, "n_layers": 168},
@@ -37,6 +39,7 @@ MCT_CONFIG = {
             "n_layers": 187,
         },
         "classify": {"layer_names": [], "weights_memory": np.inf, "n_layers": 73},
+        "segment": {"layer_names": ["sub", "mul", "add_6", "cat_18"], "weights_memory": 2580060.0, "n_layers": 195},
     },
 }
 
@@ -92,6 +95,8 @@ class FXModel(torch.nn.Module):
                 )
             if type(m) is Pose:
                 m.forward = types.MethodType(pose_forward, m)  # bind method to Detect
+            if type(m) is Segment:
+                m.forward = types.MethodType(segment_forward, m)  # bind method to Detect
             x = m(x)  # run
             y.append(x)  # save output
         return x
@@ -114,8 +119,17 @@ def pose_forward(self, x: list[torch.Tensor]) -> tuple[torch.Tensor, torch.Tenso
     return (*x, pred_kpt.permute(0, 2, 1))
 
 
+def segment_forward(self, x: list[torch.Tensor]) -> tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
+    """Forward pass for imx segmentation."""
+    p = self.proto(x[0])  # mask protos
+    bs = p.shape[0]  # batch size
+    mc = torch.cat([self.cv4[i](x[i]).view(bs, self.nm, -1) for i in range(self.nl)], 2)  # mask coefficients
+    x = Detect.forward(self, x)
+    return (*x, mc.transpose(1, 2), p)
+
+
 class NMSWrapper(torch.nn.Module):
-    """Wrap PyTorch Module with multiclass_nms layer from sony_custom_layers."""
+    """Wrap PyTorch Module with multiclass_nms layer from edge-mdt-cl."""
 
     def __init__(
         self,
@@ -143,7 +157,7 @@ class NMSWrapper(torch.nn.Module):
 
     def forward(self, images):
         """Forward pass with model inference and NMS post-processing."""
-        from sony_custom_layers.pytorch import multiclass_nms_with_indices
+        from edgemdt_cl.pytorch.nms.nms_with_indices import multiclass_nms_with_indices
 
         # model inference
         outputs = self.model(images)
@@ -159,6 +173,10 @@ class NMSWrapper(torch.nn.Module):
             kpts = outputs[2]  # (bs, max_detections, kpts 17*3)
             out_kpts = torch.gather(kpts, 1, nms_outputs.indices.unsqueeze(-1).expand(-1, -1, kpts.size(-1)))
             return nms_outputs.boxes, nms_outputs.scores, nms_outputs.labels, out_kpts
+        if self.task == "segment":
+            mc, proto = outputs[2], outputs[3]
+            out_mc = torch.gather(mc, 1, nms_outputs.indices.unsqueeze(-1).expand(-1, -1, mc.size(-1)))
+            return nms_outputs.boxes, nms_outputs.scores, nms_outputs.labels, out_mc, proto
         return nms_outputs.boxes, nms_outputs.scores, nms_outputs.labels, nms_outputs.n_valid
 
 
@@ -202,7 +220,7 @@ def torch2imx(
         >>> path, _ = export_imx(model, "model.imx", conf=0.25, iou=0.45, max_det=300)
 
     Notes:
-        - Requires model_compression_toolkit, onnx, edgemdt_tpc, and sony_custom_layers packages
+        - Requires model_compression_toolkit, onnx, edgemdt_tpc, and edge-mdt-cl packages
         - Only supports YOLOv8n and YOLO11n models (detection and pose tasks)
         - Output includes quantized ONNX model, IMX binary, and labels.txt file
     """
@@ -218,6 +236,7 @@ def torch2imx(
             img = img / 255.0
             yield [img]
 
+    # NOTE: need tpc_version to be "4.0" for IMX500 Pose estimation models
     tpc = get_target_platform_capabilities(tpc_version="4.0", device_type="imx500")
 
     bit_cfg = mct.core.BitWidthConfig()
@@ -271,9 +290,11 @@ def torch2imx(
     f = Path(str(file).replace(file.suffix, "_imx_model"))
     f.mkdir(exist_ok=True)
     onnx_model = f / Path(str(file.name).replace(file.suffix, "_imx.onnx"))  # js dir
-    mct.exporter.pytorch_export_model(
-        model=quant_model, save_model_path=onnx_model, repr_dataset=representative_dataset_gen
-    )
+
+    with onnx_export_patch():
+        mct.exporter.pytorch_export_model(
+            model=quant_model, save_model_path=onnx_model, repr_dataset=representative_dataset_gen
+        )
 
     model_onnx = onnx.load(onnx_model)  # load onnx model
     for k, v in metadata.items():

{dgenerate_ultralytics_headless-8.3.234 → dgenerate_ultralytics_headless-8.3.235}/ultralytics/utils/patches.py

@@ -159,6 +159,25 @@ def arange_patch(args):
         yield
 
 
+@contextmanager
+def onnx_export_patch():
+    """Workaround for ONNX export issues in PyTorch 2.9+ with Dynamo enabled."""
+    from ultralytics.utils.torch_utils import TORCH_2_9
+
+    if TORCH_2_9:
+        func = torch.onnx.export
+
+        def torch_export(*args, **kwargs):
+            """Return a 1-D tensor of size with values from the interval and common difference."""
+            return func(*args, **kwargs, dynamo=False)  # cast to dtype instead of passing dtype
+
+        torch.onnx.export = torch_export  # patch
+        yield
+        torch.onnx.export = func  # unpatch
+    else:
+        yield
+
+
 @contextmanager
 def override_configs(args, overrides: dict[str, Any] | None = None):
     """Context manager to temporarily override configurations in args.