ultralytics 8.3.132__tar.gz → 8.3.134__tar.gz

{ultralytics-8.3.132/ultralytics.egg-info → ultralytics-8.3.134}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ultralytics
-Version: 8.3.132
+Version: 8.3.134
 Summary: Ultralytics YOLO 🚀 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification.
 Author-email: Glenn Jocher <glenn.jocher@ultralytics.com>, Jing Qiu <jing.qiu@ultralytics.com>
 Maintainer-email: Ultralytics <hello@ultralytics.com>
@@ -59,7 +59,7 @@ Requires-Dist: mkdocstrings[python]; extra == "dev"
 Requires-Dist: mkdocs-ultralytics-plugin>=0.1.17; extra == "dev"
 Requires-Dist: mkdocs-macros-plugin>=1.0.5; extra == "dev"
 Provides-Extra: export
-Requires-Dist: onnx>=1.12.0; extra == "export"
+Requires-Dist: onnx<1.18.0,>=1.12.0; extra == "export"
 Requires-Dist: coremltools>=8.0; (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"

{ultralytics-8.3.132 → ultralytics-8.3.134}/pyproject.toml

@@ -93,7 +93,7 @@ dev = [
     "mkdocs-macros-plugin>=1.0.5"  # duplicating content (i.e. export tables) in multiple places
 ]
 export = [
-    "onnx>=1.12.0", # ONNX export
+    "onnx>=1.12.0,<1.18.0", # ONNX export
     "coremltools>=8.0; 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

{ultralytics-8.3.132 → ultralytics-8.3.134}/tests/test_cli.py

@@ -61,6 +61,7 @@ def test_rtdetr(task: str = "detect", model: str = "yolov8n-rtdetr.yaml", data:
     if TORCH_1_9:
         weights = WEIGHTS_DIR / "rtdetr-l.pt"
         run(f"yolo predict {task} model={weights} source={ASSETS / 'bus.jpg'} imgsz=160 save save_crop save_txt")
+        run(f"yolo train {task} model={weights} epochs=1 imgsz=160 cache=disk data=coco8.yaml")
 
 
 @pytest.mark.skipif(checks.IS_PYTHON_3_12, reason="MobileSAM with CLIP is not supported in Python 3.12")
@@ -126,3 +127,12 @@ def test_train_gpu(task: str, model: str, data: str) -> None:
     """Test YOLO training on GPU(s) for various tasks and models."""
     run(f"yolo train {task} model={model} data={data} imgsz=32 epochs=1 device=0")  # single GPU
     run(f"yolo train {task} model={model} data={data} imgsz=32 epochs=1 device=0,1")  # multi GPU
+
+
+@pytest.mark.parametrize(
+    "solution",
+    ["count", "blur", "workout", "heatmap", "isegment", "visioneye", "speed", "queue", "analytics", "trackzone"],
+)
+def test_solutions(solution: str) -> None:
+    """Test yolo solutions command-line modes."""
+    run(f"yolo solutions {solution} verbose=False")

{ultralytics-8.3.132 → ultralytics-8.3.134}/tests/test_cuda.py

@@ -9,7 +9,7 @@ import torch
 from tests import CUDA_DEVICE_COUNT, CUDA_IS_AVAILABLE, MODEL, SOURCE
 from ultralytics import YOLO
 from ultralytics.cfg import TASK2DATA, TASK2MODEL, TASKS
-from ultralytics.utils import ASSETS, WEIGHTS_DIR
+from ultralytics.utils import ASSETS, IS_JETSON, WEIGHTS_DIR
 from ultralytics.utils.autodevice import GPUInfo
 from ultralytics.utils.checks import check_amp
 from ultralytics.utils.torch_utils import TORCH_1_13
@@ -17,11 +17,14 @@ from ultralytics.utils.torch_utils import TORCH_1_13
 # Try to find idle devices if CUDA is available
 DEVICES = []
 if CUDA_IS_AVAILABLE:
-    gpu_info = GPUInfo()
-    gpu_info.print_status()
-    idle_gpus = gpu_info.select_idle_gpu(count=2, min_memory_mb=2048)
-    if idle_gpus:
-        DEVICES = idle_gpus
+    if IS_JETSON:
+        DEVICES = [0]  # NVIDIA Jetson only has one GPU and does not fully support pynvml library
+    else:
+        gpu_info = GPUInfo()
+        gpu_info.print_status()
+        idle_gpus = gpu_info.select_idle_gpu(count=2, min_memory_mb=2048)
+        if idle_gpus:
+            DEVICES = idle_gpus
 
 
 def test_checks():
@@ -38,6 +41,7 @@ def test_amp():
 
 
 @pytest.mark.slow
+# @pytest.mark.skipif(IS_JETSON, reason="Temporary disable ONNX for Jetson")
 @pytest.mark.skipif(not DEVICES, reason="No CUDA devices available")
 @pytest.mark.parametrize(
     "task, dynamic, int8, half, batch, simplify, nms",
@@ -49,7 +53,7 @@ def test_amp():
         if not (
             (int8 and half)
             or (task == "classify" and nms)
-            or (task == "obb" and nms and not TORCH_1_13)
+            or (task == "obb" and nms and (not TORCH_1_13 or IS_JETSON))  # obb nms fails on NVIDIA Jetson
             or (simplify and dynamic)  # onnxslim is slow when dynamic=True
         )
     ],
@@ -110,9 +114,11 @@ def test_train():
 
     device = tuple(DEVICES) if len(DEVICES) > 1 else DEVICES[0]
     results = YOLO(MODEL).train(data="coco8.yaml", imgsz=64, epochs=1, device=device)  # requires imgsz>=64
-    visible = eval(os.environ["CUDA_VISIBLE_DEVICES"])
-    assert visible == device, f"Passed GPUs '{device}', but used GPUs '{visible}'"
-    assert results is (None if len(DEVICES) > 1 else not None)  # DDP returns None, single-GPU returns metrics
+    # NVIDIA Jetson only has one GPU and therefore skipping checks
+    if not IS_JETSON:
+        visible = eval(os.environ["CUDA_VISIBLE_DEVICES"])
+        assert visible == device, f"Passed GPUs '{device}', but used GPUs '{visible}'"
+        assert results is (None if len(DEVICES) > 1 else not None)  # DDP returns None, single-GPU returns metrics
 
 
 @pytest.mark.slow

{ultralytics-8.3.132 → ultralytics-8.3.134}/tests/test_python.py

@@ -271,10 +271,12 @@ def test_results(model):
         r = r.to(device="cpu", dtype=torch.float32)
         r.save_txt(txt_file=TMP / "runs/tests/label.txt", save_conf=True)
         r.save_crop(save_dir=TMP / "runs/tests/crops/")
-        r.to_json(normalize=True)
-        r.to_df(decimals=3)
+        r.to_df(decimals=3)  # Align to_ methods: https://docs.ultralytics.com/modes/predict/#working-with-results
         r.to_csv()
         r.to_xml()
+        r.to_html()
+        r.to_json(normalize=True)
+        r.to_sql()
         r.plot(pil=True, save=True, filename=TMP / "results_plot_save.jpg")
         r.plot(conf=True, boxes=True)
         print(r, len(r), r.path)  # print after methods

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/__init__.py

@@ -1,6 +1,6 @@
 # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
 
-__version__ = "8.3.132"
+__version__ = "8.3.134"
 
 import os
 

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/data/augment.py

@@ -1170,6 +1170,8 @@ class RandomPerspective:
                 img = cv2.warpPerspective(img, M, dsize=self.size, borderValue=(114, 114, 114))
             else:  # affine
                 img = cv2.warpAffine(img, M[:2], dsize=self.size, borderValue=(114, 114, 114))
+            if img.ndim == 2:
+                img = img[..., None]
         return img, M, s
 
     def apply_bboxes(self, bboxes, M):
@@ -1824,6 +1826,8 @@ class CopyPaste(BaseMixTransform):
             cv2.drawContours(im_new, instances2.segments[[j]].astype(np.int32), -1, (1, 1, 1), cv2.FILLED)
 
         result = labels2.get("img", cv2.flip(im, 1))  # augment segments
+        if result.ndim == 2:  # cv2.flip would eliminate the last dimension for grayscale images
+            result = result[..., None]
         i = im_new.astype(bool)
         im[i] = result[i]
 

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/data/build.py

@@ -244,9 +244,9 @@ def load_inference_source(source=None, batch=1, vid_stride=1, buffer=False, chan
     elif in_memory:
         dataset = source
     elif stream:
-        dataset = LoadStreams(source, vid_stride=vid_stride, buffer=buffer)
+        dataset = LoadStreams(source, vid_stride=vid_stride, buffer=buffer, channels=channels)
     elif screenshot:
-        dataset = LoadScreenshots(source)
+        dataset = LoadScreenshots(source, channels=channels)
     elif from_img:
         dataset = LoadPilAndNumpy(source, channels=channels)
     else:

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/data/dataset.py

@@ -184,7 +184,9 @@ class YOLODataset(BaseDataset):
         [cache.pop(k) for k in ("hash", "version", "msgs")]  # remove items
         labels = cache["labels"]
         if not labels:
-            LOGGER.warning(f"No images found in {cache_path}, training may not work correctly. {HELP_URL}")
+            raise RuntimeError(
+                f"No valid images found in {cache_path}. Images with incorrectly formatted labels are ignored. {HELP_URL}"
+            )
         self.im_files = [lb["im_file"] for lb in labels]  # update im_files
 
         # Check if the dataset is all boxes or all segments
@@ -199,7 +201,7 @@ class YOLODataset(BaseDataset):
             for lb in labels:
                 lb["segments"] = []
         if len_cls == 0:
-            LOGGER.warning(f"No labels found in {cache_path}, training may not work correctly. {HELP_URL}")
+            LOGGER.warning(f"Labels are missing or empty in {cache_path}, training may not work correctly. {HELP_URL}")
         return labels
 
     def build_transforms(self, hyp=None):

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/data/loaders.py

@@ -68,6 +68,7 @@ class LoadStreams:
         shape (List[Tuple[int, int, int]]): List of shapes for each stream.
         caps (List[cv2.VideoCapture]): List of cv2.VideoCapture objects for each stream.
         bs (int): Batch size for processing.
+        cv2_flag (int): OpenCV flag for image reading (grayscale or RGB).
 
     Methods:
         update: Read stream frames in daemon thread.
@@ -89,13 +90,14 @@ class LoadStreams:
         - The class implements a buffer system to manage frame storage and retrieval.
     """
 
-    def __init__(self, sources="file.streams", vid_stride=1, buffer=False):
+    def __init__(self, sources="file.streams", vid_stride=1, buffer=False, channels=3):
         """Initialize stream loader for multiple video sources, supporting various stream types."""
         torch.backends.cudnn.benchmark = True  # faster for fixed-size inference
         self.buffer = buffer  # buffer input streams
         self.running = True  # running flag for Thread
         self.mode = "stream"
         self.vid_stride = vid_stride  # video frame-rate stride
+        self.cv2_flag = cv2.IMREAD_GRAYSCALE if channels == 1 else cv2.IMREAD_COLOR  # grayscale or RGB
 
         sources = Path(sources).read_text().rsplit() if os.path.isfile(sources) else [sources]
         n = len(sources)
@@ -131,6 +133,7 @@ class LoadStreams:
             self.fps[i] = max((fps if math.isfinite(fps) else 0) % 100, 0) or 30  # 30 FPS fallback
 
             success, im = self.caps[i].read()  # guarantee first frame
+            im = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY)[..., None] if self.cv2_flag == cv2.IMREAD_GRAYSCALE else im
             if not success or im is None:
                 raise ConnectionError(f"{st}Failed to read images from {s}")
             self.imgs[i].append(im)
@@ -149,6 +152,9 @@ class LoadStreams:
                 cap.grab()  # .read() = .grab() followed by .retrieve()
                 if n % self.vid_stride == 0:
                     success, im = cap.retrieve()
+                    im = (
+                        cv2.cvtColor(im, cv2.COLOR_BGR2GRAY)[..., None] if self.cv2_flag == cv2.IMREAD_GRAYSCALE else im
+                    )
                     if not success:
                         im = np.zeros(self.shape[i], dtype=np.uint8)
                         LOGGER.warning("Video stream unresponsive, please check your IP camera connection.")
@@ -230,6 +236,7 @@ class LoadScreenshots:
         bs (int): Batch size, set to 1.
         fps (int): Frames per second, set to 30.
         monitor (Dict[str, int]): Monitor configuration details.
+        cv2_flag (int): OpenCV flag for image reading (grayscale or RGB).
 
     Methods:
         __iter__: Returns an iterator object.
@@ -241,7 +248,7 @@ class LoadScreenshots:
         ...     print(f"Captured frame: {im.shape}")
     """
 
-    def __init__(self, source):
+    def __init__(self, source, channels=3):
         """Initialize screenshot capture with specified screen and region parameters."""
         check_requirements("mss")
         import mss  # noqa
@@ -259,6 +266,7 @@ class LoadScreenshots:
         self.sct = mss.mss()
         self.bs = 1
         self.fps = 30
+        self.cv2_flag = cv2.IMREAD_GRAYSCALE if channels == 1 else cv2.IMREAD_COLOR  # grayscale or RGB
 
         # Parse monitor shape
         monitor = self.sct.monitors[self.screen]
@@ -275,6 +283,7 @@ class LoadScreenshots:
     def __next__(self):
         """Captures and returns the next screenshot as a numpy array using the mss library."""
         im0 = np.asarray(self.sct.grab(self.monitor))[:, :, :3]  # BGRA to BGR
+        im0 = cv2.cvtColor(im0, cv2.COLOR_BGR2GRAY)[..., None] if self.cv2_flag == cv2.IMREAD_GRAYSCALE else im0
         s = f"screen {self.screen} (LTWH): {self.left},{self.top},{self.width},{self.height}: "
 
         self.frame += 1
@@ -395,6 +404,11 @@ class LoadImagesAndVideos:
 
                 if success:
                     success, im0 = self.cap.retrieve()
+                    im0 = (
+                        cv2.cvtColor(im0, cv2.COLOR_BGR2GRAY)[..., None]
+                        if self.cv2_flag == cv2.IMREAD_GRAYSCALE
+                        else im0
+                    )
                     if success:
                         self.frame += 1
                         paths.append(path)
@@ -497,6 +511,8 @@ class LoadPilAndNumpy:
             # adding new axis if it's grayscale, and converting to BGR if it's RGB
             im = im[..., None] if flag == "L" else im[..., ::-1]
             im = np.ascontiguousarray(im)  # contiguous
+        elif im.ndim == 2:  # grayscale in numpy form
+            im = im[..., None]
         return im
 
     def __len__(self):

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/data/utils.py

@@ -424,8 +424,8 @@ def check_det_dataset(dataset, autodownload=True):
 
     # Resolve paths
     path = Path(extract_dir or data.get("path") or Path(data.get("yaml_file", "")).parent)  # dataset root
-    if not path.is_absolute():
-        path = (DATASETS_DIR / path).resolve()
+    if not path.exists() and not path.is_absolute():
+        path = (DATASETS_DIR / path).resolve()  # path relative to DATASETS_DIR
 
     # Set paths
     data["path"] = path  # download scripts

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/engine/exporter.py

@@ -89,6 +89,7 @@ from ultralytics.utils import (
     MACOS_VERSION,
     RKNN_CHIPS,
     ROOT,
+    SETTINGS,
     WINDOWS,
     YAML,
     callbacks,
@@ -106,7 +107,7 @@ from ultralytics.utils.downloads import attempt_download_asset, get_github_asset
 from ultralytics.utils.export import export_engine, export_onnx
 from ultralytics.utils.files import file_size, spaces_in_path
 from ultralytics.utils.ops import Profile, nms_rotated
-from ultralytics.utils.torch_utils import TORCH_1_13, get_latest_opset, select_device
+from ultralytics.utils.torch_utils import TORCH_1_13, get_cpu_info, get_latest_opset, select_device
 
 
 def export_formats():
@@ -141,7 +142,7 @@ def export_formats():
         ["MNN", "mnn", ".mnn", True, True, ["batch", "half", "int8"]],
         ["NCNN", "ncnn", "_ncnn_model", True, True, ["batch", "half"]],
         ["IMX", "imx", "_imx_model", True, True, ["int8", "fraction"]],
-        ["RKNN", "rknn", "_rknn_model", False, False, ["batch", "name", "int8"]],
+        ["RKNN", "rknn", "_rknn_model", False, False, ["batch", "name"]],
     ]
     return dict(zip(["Format", "Argument", "Suffix", "CPU", "GPU", "Arguments"], zip(*x)))
 
@@ -344,7 +345,6 @@ class Exporter:
                 "See https://docs.ultralytics.com/models/yolo-world for details."
             )
             model.clip_model = None  # openvino int8 export error: https://github.com/ultralytics/ultralytics/pull/18445
-
         if self.args.int8 and not self.args.data:
             self.args.data = DEFAULT_CFG.data or TASK2DATA[getattr(model, "task", "detect")]  # assign default data
             LOGGER.warning(
@@ -352,6 +352,14 @@ class Exporter:
             )
         if tfjs and (ARM64 and LINUX):
             raise SystemError("TF.js exports are not currently supported on ARM64 Linux")
+        # Recommend OpenVINO if export and Intel CPU
+        if SETTINGS.get("openvino_msg"):
+            if "intel" in get_cpu_info().lower():
+                LOGGER.info(
+                    "💡 ProTip: Export to OpenVINO format for best performance on Intel CPUs."
+                    " Learn more at https://docs.ultralytics.com/integrations/openvino/"
+                )
+            SETTINGS["openvino_msg"] = False
 
         # Input
         im = torch.zeros(self.args.batch, model.yaml.get("channels", 3), *self.imgsz).to(self.device)
@@ -547,7 +555,7 @@ class Exporter:
     @try_export
     def export_onnx(self, prefix=colorstr("ONNX:")):
         """YOLO ONNX export."""
-        requirements = ["onnx>=1.12.0"]
+        requirements = ["onnx>=1.12.0,<1.18.0"]
         if self.args.simplify:
             requirements += ["onnxslim>=0.1.46", "onnxruntime" + ("-gpu" if torch.cuda.is_available() else "")]
         check_requirements(requirements)
@@ -1113,8 +1121,8 @@ class Exporter:
         rknn = RKNN(verbose=False)
         rknn.config(mean_values=[[0, 0, 0]], std_values=[[255, 255, 255]], target_platform=self.args.name)
         rknn.load_onnx(model=f)
-        rknn.build(do_quantization=self.args.int8)
-        f = f.replace(".onnx", f"-{self.args.name}-int8.rknn" if self.args.int8 else f"-{self.args.name}-fp16.rknn")
+        rknn.build(do_quantization=False)  # TODO: Add quantization support
+        f = f.replace(".onnx", f"-{self.args.name}.rknn")
         rknn.export_rknn(f"{export_path / f}")
         YAML.save(export_path / "metadata.yaml", self.metadata)
         return export_path, None

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/engine/model.py

@@ -529,7 +529,7 @@ class Model(torch.nn.Module):
             - For SAM-type models, 'prompts' can be passed as a keyword argument.
         """
         if source is None:
-            source = ASSETS
+            source = "https://ultralytics.com/images/boats.jpg" if self.task == "obb" else ASSETS
             LOGGER.warning(f"'source' is missing. Using 'source={source}'.")
 
         is_cli = (ARGV[0].endswith("yolo") or ARGV[0].endswith("ultralytics")) and any(

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/engine/results.py

@@ -1032,7 +1032,7 @@ class Results(SimpleClass):
         conn.commit()
         conn.close()
 
-        LOGGER.info(f"✅ Detection results successfully written to SQL table '{table_name}' in database '{db_path}'.")
+        LOGGER.info(f"Detection results successfully written to SQL table '{table_name}' in database '{db_path}'.")
 
 
 class Boxes(BaseTensor):

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/models/yolo/detect/predict.py

@@ -51,7 +51,7 @@ class DetectionPredictor(BasePredictor):
             >>> results = predictor.predict("path/to/image.jpg")
             >>> processed_results = predictor.postprocess(preds, img, orig_imgs)
         """
-        save_feats = getattr(self, "save_feats", False)
+        save_feats = getattr(self, "_feats", None) is not None
         preds = ops.non_max_suppression(
             preds,
             self.args.conf,

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/nn/tasks.py

@@ -284,13 +284,15 @@ class BaseModel(torch.nn.Module):
         updated_csd = intersect_dicts(csd, self.state_dict())  # intersect
         self.load_state_dict(updated_csd, strict=False)  # load
         len_updated_csd = len(updated_csd)
-        first_conv = "model.0.conv.weight"
-        if first_conv not in updated_csd:  # mostly used to boost multi-channel training
-            c1, c2, h, w = self.state_dict()[first_conv].shape
+        first_conv = "model.0.conv.weight"  # hard-coded to yolo models for now
+        # mostly used to boost multi-channel training
+        state_dict = self.state_dict()
+        if first_conv not in updated_csd and first_conv in state_dict:
+            c1, c2, h, w = state_dict[first_conv].shape
             cc1, cc2, ch, cw = csd[first_conv].shape
             if ch == h and cw == w:
                 c1, c2 = min(c1, cc1), min(c2, cc2)
-                self.state_dict()[first_conv][:c1, :c2] = csd[first_conv][:c1, :c2]
+                state_dict[first_conv][:c1, :c2] = csd[first_conv][:c1, :c2]
                 len_updated_csd += 1
         if verbose:
             LOGGER.info(f"Transferred {len_updated_csd}/{len(self.model.state_dict())} items from pretrained weights")

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/solutions/ai_gym.py

@@ -1,5 +1,7 @@
 # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
 
+from collections import defaultdict
+
 from ultralytics.solutions.solutions import BaseSolution, SolutionAnnotator, SolutionResults
 
 
@@ -11,10 +13,7 @@ class AIGym(BaseSolution):
     repetitions of exercises based on predefined angle thresholds for up and down positions.
 
     Attributes:
-        count (List[int]): Repetition counts for each detected person.
-        angle (List[float]): Current angle of the tracked body part for each person.
-        stage (List[str]): Current exercise stage ('up', 'down', or '-') for each person.
-        initial_stage (str | None): Initial stage of the exercise.
+        states (Dict[float, int, str]): Stores per-track angle, count, and stage for workout monitoring.
         up_angle (float): Angle threshold for considering the 'up' position of an exercise.
         down_angle (float): Angle threshold for considering the 'down' position of an exercise.
         kpts (List[int]): Indices of keypoints used for angle calculation.
@@ -41,12 +40,9 @@ class AIGym(BaseSolution):
         """
         kwargs["model"] = kwargs.get("model", "yolo11n-pose.pt")
         super().__init__(**kwargs)
-        self.count = []  # List for counts, necessary where there are multiple objects in frame
-        self.angle = []  # List for angle, necessary where there are multiple objects in frame
-        self.stage = []  # List for stage, necessary where there are multiple objects in frame
+        self.states = defaultdict(lambda: {"angle": 0, "count": 0, "stage": "-"})  # Dict for count, angle and stage
 
         # Extract details from CFG single time for usage later
-        self.initial_stage = None
         self.up_angle = float(self.CFG["up_angle"])  # Pose up predefined angle to consider up pose
         self.down_angle = float(self.CFG["down_angle"])  # Pose down predefined angle to consider down pose
         self.kpts = self.CFG["kpts"]  # User selected kpts of workouts storage for further usage
@@ -81,33 +77,30 @@ class AIGym(BaseSolution):
         tracks = self.tracks[0]
 
         if tracks.boxes.id is not None:
-            if len(tracks) > len(self.count):  # Add new entries for newly detected people
-                new_human = len(tracks) - len(self.count)
-                self.angle += [0] * new_human
-                self.count += [0] * new_human
-                self.stage += ["-"] * new_human
-
-            # Enumerate over keypoints
-            for ind, k in enumerate(reversed(tracks.keypoints.data)):
+            track_ids = tracks.boxes.id.cpu().tolist()
+            kpt_data = tracks.keypoints.data.cpu()  # Avoid repeated .cpu() calls
+
+            for i, k in enumerate(kpt_data):
+                track_id = int(track_ids[i])  # get track id
+                state = self.states[track_id]  # get state details
                 # Get keypoints and estimate the angle
-                kpts = [k[int(self.kpts[i])].cpu() for i in range(3)]
-                self.angle[ind] = annotator.estimate_pose_angle(*kpts)
+                state["angle"] = annotator.estimate_pose_angle(*[k[int(idx)] for idx in self.kpts])
                 annotator.draw_specific_kpts(k, self.kpts, radius=self.line_width * 3)
 
                 # Determine stage and count logic based on angle thresholds
-                if self.angle[ind] < self.down_angle:
-                    if self.stage[ind] == "up":
-                        self.count[ind] += 1
-                    self.stage[ind] = "down"
-                elif self.angle[ind] > self.up_angle:
-                    self.stage[ind] = "up"
+                if state["angle"] < self.down_angle:
+                    if state["stage"] == "up":
+                        state["count"] += 1
+                    state["stage"] = "down"
+                elif state["angle"] > self.up_angle:
+                    state["stage"] = "up"
 
                 # Display angle, count, and stage text
                 if self.show_labels:
                     annotator.plot_angle_and_count_and_stage(
-                        angle_text=self.angle[ind],  # angle text for display
-                        count_text=self.count[ind],  # count text for workouts
-                        stage_text=self.stage[ind],  # stage position text
+                        angle_text=state["angle"],  # angle text for display
+                        count_text=state["count"],  # count text for workouts
+                        stage_text=state["stage"],  # stage position text
                         center_kpt=k[int(self.kpts[1])],  # center keypoint for display
                     )
         plot_im = annotator.result()
@@ -116,8 +109,8 @@ class AIGym(BaseSolution):
         # Return SolutionResults
         return SolutionResults(
             plot_im=plot_im,
-            workout_count=self.count,
-            workout_stage=self.stage,
-            workout_angle=self.angle,
+            workout_count=[v["count"] for v in self.states.values()],
+            workout_stage=[v["stage"] for v in self.states.values()],
+            workout_angle=[v["angle"] for v in self.states.values()],
             total_tracks=len(self.track_ids),
         )

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/solutions/heatmap.py

@@ -99,7 +99,6 @@ class Heatmap(ObjectCounter):
             if self.region is not None:
                 self.annotator.draw_region(reg_pts=self.region, color=(104, 0, 123), thickness=self.line_width * 2)
                 self.store_tracking_history(track_id, box)  # Store track history
-                self.store_classwise_counts(cls)  # Store classwise counts in dict
                 # Get previous position if available
                 prev_position = None
                 if len(self.track_history[track_id]) > 1:
@@ -123,6 +122,6 @@ class Heatmap(ObjectCounter):
             plot_im=plot_im,
             in_count=self.in_count,
             out_count=self.out_count,
-            classwise_count=self.classwise_counts,
+            classwise_count=dict(self.classwise_counts),
             total_tracks=len(self.track_ids),
         )

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/solutions/object_counter.py

@@ -1,5 +1,7 @@
 # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
 
+from collections import defaultdict
+
 from ultralytics.solutions.solutions import BaseSolution, SolutionAnnotator, SolutionResults
 from ultralytics.utils.plotting import colors
 
@@ -22,7 +24,6 @@ class ObjectCounter(BaseSolution):
 
     Methods:
         count_objects: Counts objects within a polygonal or linear region.
-        store_classwise_counts: Initializes class-wise counts if not already present.
         display_counts: Displays object counts on the frame.
         process: Processes input data (frames or object tracks) and updates counts.
 
@@ -40,7 +41,7 @@ class ObjectCounter(BaseSolution):
         self.in_count = 0  # Counter for objects moving inward
         self.out_count = 0  # Counter for objects moving outward
         self.counted_ids = []  # List of IDs of objects that have been counted
-        self.classwise_counts = {}  # Dictionary for counts, categorized by object class
+        self.classwise_counts = defaultdict(lambda: {"IN": 0, "OUT": 0})  # Dictionary for counts, categorized by class
         self.region_initialized = False  # Flag indicating whether the region has been initialized
 
         self.show_in = self.CFG["show_in"]
@@ -110,22 +111,6 @@ class ObjectCounter(BaseSolution):
                     self.classwise_counts[self.names[cls]]["OUT"] += 1
                 self.counted_ids.append(track_id)
 
-    def store_classwise_counts(self, cls):
-        """
-        Initialize class-wise counts for a specific object class if not already present.
-
-        Args:
-            cls (int): Class index for classwise count updates.
-
-        Examples:
-            >>> counter = ObjectCounter()
-            >>> counter.store_classwise_counts(0)  # Initialize counts for class index 0
-            >>> print(counter.classwise_counts)
-            {'person': {'IN': 0, 'OUT': 0}}
-        """
-        if self.names[cls] not in self.classwise_counts:
-            self.classwise_counts[self.names[cls]] = {"IN": 0, "OUT": 0}
-
     def display_counts(self, plot_im):
         """
         Display object counts on the input image or frame.
@@ -189,7 +174,6 @@ class ObjectCounter(BaseSolution):
                 box, label=self.adjust_box_label(cls, conf, track_id), color=colors(cls, True), rotated=is_obb
             )
             self.store_tracking_history(track_id, box, is_obb=is_obb)  # Store track history
-            self.store_classwise_counts(cls)  # Store classwise counts in dict
 
             # Store previous position of track for object counting
             prev_position = None
@@ -206,6 +190,6 @@ class ObjectCounter(BaseSolution):
             plot_im=plot_im,
             in_count=self.in_count,
             out_count=self.out_count,
-            classwise_count=self.classwise_counts,
+            classwise_count=dict(self.classwise_counts),
             total_tracks=len(self.track_ids),
         )

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/solutions/security_alarm.py

@@ -110,7 +110,7 @@ class SecurityAlarm(BaseSolution):
         # Send the email
         try:
             self.server.send_message(message)
-            LOGGER.info("✅ Email sent successfully!")
+            LOGGER.info("Email sent successfully!")
         except Exception as e:
             LOGGER.error(f"Failed to send email: {e}")
 

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/trackers/byte_tracker.py

@@ -330,7 +330,11 @@ class BYTETracker:
         # Predict the current location with KF
         self.multi_predict(strack_pool)
         if hasattr(self, "gmc") and img is not None:
-            warp = self.gmc.apply(img, dets)
+            # use try-except here to bypass errors from gmc module
+            try:
+                warp = self.gmc.apply(img, dets)
+            except Exception:
+                warp = np.eye(2, 3)
             STrack.multi_gmc(strack_pool, warp)
             STrack.multi_gmc(unconfirmed, warp)
 

{ultralytics-8.3.132 → ultralytics-8.3.134}/ultralytics/trackers/track.py

@@ -45,7 +45,8 @@ def on_predict_start(predictor: object, persist: bool = False) -> None:
         raise AssertionError(f"Only 'bytetrack' and 'botsort' are supported for now, but got '{cfg.tracker_type}'")
 
     predictor._feats = None  # reset in case used earlier
-    predictor.save_feats = False
+    if hasattr(predictor, "_hook"):
+        predictor._hook.remove()
     if cfg.tracker_type == "botsort" and cfg.with_reid and cfg.model == "auto":
         from ultralytics.nn.modules.head import Detect
 
@@ -56,13 +57,11 @@ def on_predict_start(predictor: object, persist: bool = False) -> None:
         ):
             cfg.model = "yolo11n-cls.pt"
         else:
-            predictor.save_feats = True
-
             # Register hook to extract input of Detect layer
             def pre_hook(module, input):
-                predictor._feats = [t.clone() for t in input[0]]
+                predictor._feats = list(input[0])  # unroll to new list to avoid mutation in forward
 
-            predictor.model.model.model[-1].register_forward_pre_hook(pre_hook)
+            predictor._hook = predictor.model.model.model[-1].register_forward_pre_hook(pre_hook)
 
     trackers = []
     for _ in range(predictor.dataset.bs):