ultralytics 8.0.131 NCNN, HUB and FastSAM fixes (#3587)

ultralytics/yolo/engine/exporter.py

@@ -50,7 +50,6 @@ TensorFlow.js:
 """
 import json
 import os
-import platform
 import shutil
 import subprocess
 import time
@@ -64,7 +63,7 @@ from ultralytics.nn.autobackend import check_class_names
 from ultralytics.nn.modules import C2f, Detect, RTDETRDecoder
 from ultralytics.nn.tasks import DetectionModel, SegmentationModel
 from ultralytics.yolo.cfg import get_cfg
-from ultralytics.yolo.utils import (DEFAULT_CFG, LINUX, LOGGER, MACOS, ROOT, __version__, callbacks, colorstr,
+from ultralytics.yolo.utils import (ARM64, DEFAULT_CFG, LINUX, LOGGER, MACOS, ROOT, __version__, callbacks, colorstr,
                                     get_default_args, yaml_save)
 from ultralytics.yolo.utils.checks import check_imgsz, check_requirements, check_version
 from ultralytics.yolo.utils.downloads import attempt_download_asset, get_github_assets
@@ -72,8 +71,6 @@ from ultralytics.yolo.utils.files import file_size
 from ultralytics.yolo.utils.ops import Profile
 from ultralytics.yolo.utils.torch_utils import get_latest_opset, select_device, smart_inference_mode
 
-ARM64 = platform.machine() in ('arm64', 'aarch64')
-
 
 def export_formats():
     """YOLOv8 export formats."""
@@ -170,7 +167,8 @@ class Exporter:
             assert not self.args.dynamic, 'half=True not compatible with dynamic=True, i.e. use only one.'
         self.imgsz = check_imgsz(self.args.imgsz, stride=model.stride, min_dim=2)  # check image size
         if self.args.optimize:
-            assert self.device.type == 'cpu', '--optimize not compatible with cuda devices, i.e. use --device cpu'
+            assert not ncnn, "optimize=True not compatible with format='ncnn', i.e. use optimize=False"
+            assert self.device.type == 'cpu', "optimize=True not compatible with cuda devices, i.e. use device='cpu'"
         if edgetpu and not LINUX:
             raise SystemError('Edge TPU export only supported on Linux. See https://coral.ai/docs/edgetpu/compiler/')
 
@@ -405,7 +403,7 @@ class Exporter:
         """
         YOLOv8 NCNN export using PNNX https://github.com/pnnx/pnnx.
         """
-        check_requirements('ncnn')  # requires NCNN
+        check_requirements('git+https://github.com/Tencent/ncnn.git' if ARM64 else 'ncnn')  # requires NCNN
         import ncnn  # noqa
 
         LOGGER.info(f'\n{prefix} starting export with NCNN {ncnn.__version__}...')

ultralytics/yolo/fastsam/model.py

@@ -22,7 +22,7 @@ from .predict import FastSAMPredictor
 class FastSAM(YOLO):
 
     def __init__(self, model='FastSAM-x.pt'):
-        # Call the __init__ method of the parent class (YOLO) with the updated default model
+        """Call the __init__ method of the parent class (YOLO) with the updated default model"""
         if model == 'FastSAM.pt':
             model = 'FastSAM-x.pt'
         super().__init__(model=model)

ultralytics/yolo/fastsam/utils.py

@@ -20,11 +20,10 @@ def adjust_bboxes_to_image_border(boxes, image_shape, threshold=20):
     h, w = image_shape
 
     # Adjust boxes
-    boxes[:, 0] = torch.where(boxes[:, 0] < threshold, 0, boxes[:, 0])  # x1
-    boxes[:, 1] = torch.where(boxes[:, 1] < threshold, 0, boxes[:, 1])  # y1
-    boxes[:, 2] = torch.where(boxes[:, 2] > w - threshold, w, boxes[:, 2])  # x2
-    boxes[:, 3] = torch.where(boxes[:, 3] > h - threshold, h, boxes[:, 3])  # y2
-
+    boxes[boxes[:, 0] < threshold, 0] = 0  # x1
+    boxes[boxes[:, 1] < threshold, 1] = 0  # y1
+    boxes[boxes[:, 2] > w - threshold, 2] = w  # x2
+    boxes[boxes[:, 3] > h - threshold, 3] = h  # y2
     return boxes
 
 

ultralytics/yolo/utils/__init__.py

@@ -38,6 +38,7 @@ VERBOSE = str(os.getenv('YOLO_VERBOSE', True)).lower() == 'true'  # global verbo
 TQDM_BAR_FORMAT = '{l_bar}{bar:10}{r_bar}'  # tqdm bar format
 LOGGING_NAME = 'ultralytics'
 MACOS, LINUX, WINDOWS = (platform.system() == x for x in ['Darwin', 'Linux', 'Windows'])  # environment booleans
+ARM64 = platform.machine() in ('arm64', 'aarch64')  # ARM64 booleans
 HELP_MSG = \
     """
     Usage examples for running YOLOv8: