ultralytics 8.3.158__tar.gz → 8.3.160__tar.gz

{ultralytics-8.3.158/ultralytics.egg-info → ultralytics-8.3.160}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ultralytics
-Version: 8.3.158
+Version: 8.3.160
 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>
@@ -87,7 +87,7 @@ Dynamic: license-file
       <img width="100%" src="https://raw.githubusercontent.com/ultralytics/assets/main/yolov8/banner-yolov8.png" alt="Ultralytics YOLO banner"></a>
   </p>
 
-[中文](https://docs.ultralytics.com/zh) | [한국어](https://docs.ultralytics.com/ko) | [日本語](https://docs.ultralytics.com/ja) | [Русский](https://docs.ultralytics.com/ru) | [Deutsch](https://docs.ultralytics.com/de) | [Français](https://docs.ultralytics.com/fr) | [Español](https://docs.ultralytics.com/es) | [Português](https://docs.ultralytics.com/pt) | [Türkçe](https://docs.ultralytics.com/tr) | [Tiếng Việt](https://docs.ultralytics.com/vi) | [العربية](https://docs.ultralytics.com/ar) <br>
+[中文](https://docs.ultralytics.com/zh/) | [한국어](https://docs.ultralytics.com/ko/) | [日本語](https://docs.ultralytics.com/ja/) | [Русский](https://docs.ultralytics.com/ru/) | [Deutsch](https://docs.ultralytics.com/de/) | [Français](https://docs.ultralytics.com/fr/) | [Español](https://docs.ultralytics.com/es) | [Português](https://docs.ultralytics.com/pt/) | [Türkçe](https://docs.ultralytics.com/tr/) | [Tiếng Việt](https://docs.ultralytics.com/vi/) | [العربية](https://docs.ultralytics.com/ar/) <br>
 
 <div>
     <a href="https://github.com/ultralytics/ultralytics/actions/workflows/ci.yml"><img src="https://github.com/ultralytics/ultralytics/actions/workflows/ci.yml/badge.svg" alt="Ultralytics CI"></a>

{ultralytics-8.3.158 → ultralytics-8.3.160}/README.md

@@ -4,7 +4,7 @@
       <img width="100%" src="https://raw.githubusercontent.com/ultralytics/assets/main/yolov8/banner-yolov8.png" alt="Ultralytics YOLO banner"></a>
   </p>
 
-[中文](https://docs.ultralytics.com/zh) | [한국어](https://docs.ultralytics.com/ko) | [日本語](https://docs.ultralytics.com/ja) | [Русский](https://docs.ultralytics.com/ru) | [Deutsch](https://docs.ultralytics.com/de) | [Français](https://docs.ultralytics.com/fr) | [Español](https://docs.ultralytics.com/es) | [Português](https://docs.ultralytics.com/pt) | [Türkçe](https://docs.ultralytics.com/tr) | [Tiếng Việt](https://docs.ultralytics.com/vi) | [العربية](https://docs.ultralytics.com/ar) <br>
+[中文](https://docs.ultralytics.com/zh/) | [한국어](https://docs.ultralytics.com/ko/) | [日本語](https://docs.ultralytics.com/ja/) | [Русский](https://docs.ultralytics.com/ru/) | [Deutsch](https://docs.ultralytics.com/de/) | [Français](https://docs.ultralytics.com/fr/) | [Español](https://docs.ultralytics.com/es) | [Português](https://docs.ultralytics.com/pt/) | [Türkçe](https://docs.ultralytics.com/tr/) | [Tiếng Việt](https://docs.ultralytics.com/vi/) | [العربية](https://docs.ultralytics.com/ar/) <br>
 
 <div>
     <a href="https://github.com/ultralytics/ultralytics/actions/workflows/ci.yml"><img src="https://github.com/ultralytics/ultralytics/actions/workflows/ci.yml/badge.svg" alt="Ultralytics CI"></a>

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/__init__.py

@@ -1,6 +1,6 @@
 # Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
 
-__version__ = "8.3.158"
+__version__ = "8.3.160"
 
 import os
 

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/cfg/__init__.py

@@ -954,8 +954,6 @@ def entrypoint(debug: str = "") -> None:
         from ultralytics import YOLO
 
         model = YOLO(model, task=task)
-    if isinstance(overrides.get("pretrained"), str):
-        model.load(overrides["pretrained"])
 
     # Task Update
     if task != model.task:

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/data/augment.py

@@ -251,8 +251,7 @@ class Compose:
             >>> multiple_transforms = compose[0:2]  # Returns a Compose object with RandomFlip and RandomPerspective
         """
         assert isinstance(index, (int, list)), f"The indices should be either list or int type but got {type(index)}"
-        index = [index] if isinstance(index, int) else index
-        return Compose([self.transforms[i] for i in index])
+        return Compose([self.transforms[i] for i in index]) if isinstance(index, list) else self.transforms[index]
 
     def __setitem__(self, index: Union[list, int], value: Union[list, int]) -> None:
         """
@@ -1560,14 +1559,15 @@ class RandomFlip:
         h = 1 if instances.normalized else h
         w = 1 if instances.normalized else w
 
-        # Flip up-down
+        # WARNING: two separate if and calls to random.random() intentional for reproducibility with older versions
         if self.direction == "vertical" and random.random() < self.p:
             img = np.flipud(img)
             instances.flipud(h)
+            if self.flip_idx is not None and instances.keypoints is not None:
+                instances.keypoints = np.ascontiguousarray(instances.keypoints[:, self.flip_idx, :])
         if self.direction == "horizontal" and random.random() < self.p:
             img = np.fliplr(img)
             instances.fliplr(w)
-            # For keypoints
             if self.flip_idx is not None and instances.keypoints is not None:
                 instances.keypoints = np.ascontiguousarray(instances.keypoints[:, self.flip_idx, :])
         labels["img"] = np.ascontiguousarray(img)
@@ -2533,9 +2533,9 @@ def v8_transforms(dataset, imgsz, hyp, stretch=False):
     flip_idx = dataset.data.get("flip_idx", [])  # for keypoints augmentation
     if dataset.use_keypoints:
         kpt_shape = dataset.data.get("kpt_shape", None)
-        if len(flip_idx) == 0 and hyp.fliplr > 0.0:
-            hyp.fliplr = 0.0
-            LOGGER.warning("No 'flip_idx' array defined in data.yaml, setting augmentation 'fliplr=0.0'")
+        if len(flip_idx) == 0 and (hyp.fliplr > 0.0 or hyp.flipud > 0.0):
+            hyp.fliplr = hyp.flipud = 0.0  # both fliplr and flipud require flip_idx
+            LOGGER.warning("No 'flip_idx' array defined in data.yaml, disabling 'fliplr' and 'flipud' augmentations.")
         elif flip_idx and (len(flip_idx) != kpt_shape[0]):
             raise ValueError(f"data.yaml flip_idx={flip_idx} length must be equal to kpt_shape[0]={kpt_shape[0]}")
 
@@ -2546,7 +2546,7 @@ def v8_transforms(dataset, imgsz, hyp, stretch=False):
             CutMix(dataset, pre_transform=pre_transform, p=hyp.cutmix),
             Albumentations(p=1.0),
             RandomHSV(hgain=hyp.hsv_h, sgain=hyp.hsv_s, vgain=hyp.hsv_v),
-            RandomFlip(direction="vertical", p=hyp.flipud),
+            RandomFlip(direction="vertical", p=hyp.flipud, flip_idx=flip_idx),
             RandomFlip(direction="horizontal", p=hyp.fliplr, flip_idx=flip_idx),
         ]
     )  # transforms

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/engine/exporter.py

@@ -1152,7 +1152,9 @@ class Exporter:
         )
         if getattr(self.model, "end2end", False):
             raise ValueError("IMX export is not supported for end2end models.")
-        check_requirements(("model-compression-toolkit>=2.3.0", "sony-custom-layers>=0.3.0", "edge-mdt-tpc>=1.1.0"))
+        check_requirements(
+            ("model-compression-toolkit>=2.3.0,<2.4.1", "sony-custom-layers>=0.3.0", "edge-mdt-tpc>=1.1.0")
+        )
         check_requirements("imx500-converter[pt]>=3.16.1")  # Separate requirements for imx500-converter
 
         import model_compression_toolkit as mct
@@ -1493,7 +1495,7 @@ class NMSModel(torch.nn.Module):
         scores, classes = scores.max(dim=-1)
         self.args.max_det = min(pred.shape[1], self.args.max_det)  # in case num_anchors < max_det
         # (N, max_det, 4 coords + 1 class score + 1 class label + extra_shape).
-        out = torch.zeros(bs, self.args.max_det, boxes.shape[-1] + 2 + extra_shape, **kwargs)
+        out = torch.zeros(pred.shape[0], self.args.max_det, boxes.shape[-1] + 2 + extra_shape, **kwargs)
         for i in range(bs):
             box, cls, score, extra = boxes[i], classes[i], scores[i], extras[i]
             mask = score > self.args.conf

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/engine/model.py

@@ -777,6 +777,8 @@ class Model(torch.nn.Module):
 
         checks.check_pip_update_available()
 
+        if isinstance(kwargs.get("pretrained", None), (str, Path)):
+            self.load(kwargs["pretrained"])  # load pretrained weights if provided
         overrides = YAML.load(checks.check_yaml(kwargs["cfg"])) if kwargs.get("cfg") else self.overrides
         custom = {
             # NOTE: handle the case when 'cfg' includes 'data'.

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/engine/results.py

@@ -16,7 +16,6 @@ import torch
 from ultralytics.data.augment import LetterBox
 from ultralytics.utils import LOGGER, DataExportMixin, SimpleClass, ops
 from ultralytics.utils.plotting import Annotator, colors, save_one_box
-from ultralytics.utils.torch_utils import smart_inference_mode
 
 
 class BaseTensor(SimpleClass):
@@ -1204,7 +1203,6 @@ class Keypoints(BaseTensor):
         >>> keypoints_cpu = keypoints.cpu()  # Move keypoints to CPU
     """
 
-    @smart_inference_mode()  # avoid keypoints < conf in-place error
     def __init__(self, keypoints: Union[torch.Tensor, np.ndarray], orig_shape: Tuple[int, int]) -> None:
         """
         Initialize the Keypoints object with detection keypoints and original image dimensions.
@@ -1225,9 +1223,6 @@ class Keypoints(BaseTensor):
         """
         if keypoints.ndim == 2:
             keypoints = keypoints[None, :]
-        if keypoints.shape[2] == 3:  # x, y, conf
-            mask = keypoints[..., 2] < 0.5  # points with conf < 0.5 (not visible)
-            keypoints[..., :2][mask] = 0
         super().__init__(keypoints, orig_shape)
         self.has_visible = self.data.shape[-1] == 3
 

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/models/yolo/detect/val.py

@@ -2,7 +2,7 @@
 
 import os
 from pathlib import Path
-from typing import Any, Dict, List, Optional, Tuple
+from typing import Any, Dict, List, Optional, Tuple, Union
 
 import numpy as np
 import torch
@@ -219,6 +219,7 @@ class DetectionValidator(BaseValidator):
                 self.confusion_matrix.plot(save_dir=self.save_dir, normalize=normalize, on_plot=self.on_plot)
         self.metrics.speed = self.speed
         self.metrics.confusion_matrix = self.confusion_matrix
+        self.metrics.save_dir = self.save_dir
 
     def get_stats(self) -> Dict[str, Any]:
         """
@@ -392,38 +393,73 @@ class DetectionValidator(BaseValidator):
         Returns:
             (Dict[str, Any]): Updated statistics dictionary with COCO/LVIS evaluation results.
         """
-        if self.args.save_json and (self.is_coco or self.is_lvis) and len(self.jdict):
-            pred_json = self.save_dir / "predictions.json"  # predictions
-            anno_json = (
-                self.data["path"]
-                / "annotations"
-                / ("instances_val2017.json" if self.is_coco else f"lvis_v1_{self.args.split}.json")
-            )  # annotations
+        pred_json = self.save_dir / "predictions.json"  # predictions
+        anno_json = (
+            self.data["path"]
+            / "annotations"
+            / ("instances_val2017.json" if self.is_coco else f"lvis_v1_{self.args.split}.json")
+        )  # annotations
+        return self.coco_evaluate(stats, pred_json, anno_json)
+
+    def coco_evaluate(
+        self,
+        stats: Dict[str, Any],
+        pred_json: str,
+        anno_json: str,
+        iou_types: Union[str, List[str]] = "bbox",
+        suffix: Union[str, List[str]] = "Box",
+    ) -> Dict[str, Any]:
+        """
+        Evaluate COCO/LVIS metrics using faster-coco-eval library.
+
+        Performs evaluation using the faster-coco-eval library to compute mAP metrics
+        for object detection. Updates the provided stats dictionary with computed metrics
+        including mAP50, mAP50-95, and LVIS-specific metrics if applicable.
+
+        Args:
+            stats (Dict[str, Any]): Dictionary to store computed metrics and statistics.
+            pred_json (str | Path]): Path to JSON file containing predictions in COCO format.
+            anno_json (str | Path]): Path to JSON file containing ground truth annotations in COCO format.
+            iou_types (str | List[str]]): IoU type(s) for evaluation. Can be single string or list of strings.
+                Common values include "bbox", "segm", "keypoints". Defaults to "bbox".
+            suffix (str | List[str]]): Suffix to append to metric names in stats dictionary. Should correspond
+                to iou_types if multiple types provided. Defaults to "Box".
 
+        Returns:
+            (Dict[str, Any]): Updated stats dictionary containing the computed COCO/LVIS evaluation metrics.
+        """
+        if self.args.save_json and (self.is_coco or self.is_lvis) and len(self.jdict):
             LOGGER.info(f"\nEvaluating faster-coco-eval mAP using {pred_json} and {anno_json}...")
             try:
                 for x in pred_json, anno_json:
                     assert x.is_file(), f"{x} file not found"
+                iou_types = [iou_types] if isinstance(iou_types, str) else iou_types
+                suffix = [suffix] if isinstance(suffix, str) else suffix
                 check_requirements("faster-coco-eval>=1.6.7")
                 from faster_coco_eval import COCO, COCOeval_faster
 
                 anno = COCO(anno_json)
                 pred = anno.loadRes(pred_json)
-                val = COCOeval_faster(anno, pred, iouType="bbox", lvis_style=self.is_lvis, print_function=LOGGER.info)
-                val.params.imgIds = [int(Path(x).stem) for x in self.dataloader.dataset.im_files]  # images to eval
-                val.evaluate()
-                val.accumulate()
-                val.summarize()
+                for i, iou_type in enumerate(iou_types):
+                    val = COCOeval_faster(
+                        anno, pred, iouType=iou_type, lvis_style=self.is_lvis, print_function=LOGGER.info
+                    )
+                    val.params.imgIds = [int(Path(x).stem) for x in self.dataloader.dataset.im_files]  # images to eval
+                    val.evaluate()
+                    val.accumulate()
+                    val.summarize()
+
+                    # update mAP50-95 and mAP50
+                    stats[f"metrics/mAP50({suffix[i][0]})"] = val.stats_as_dict["AP_all"]
+                    stats[f"metrics/mAP50-95({suffix[i][0]})"] = val.stats_as_dict["AP_50"]
 
-                # update mAP50-95 and mAP50
-                stats[self.metrics.keys[-1]] = val.stats_as_dict["AP_all"]
-                stats[self.metrics.keys[-2]] = val.stats_as_dict["AP_50"]
+                    if self.is_lvis:
+                        stats[f"metrics/APr({suffix[i][0]})"] = val.stats_as_dict["APr"]
+                        stats[f"metrics/APc({suffix[i][0]})"] = val.stats_as_dict["APc"]
+                        stats[f"metrics/APf({suffix[i][0]})"] = val.stats_as_dict["APf"]
 
                 if self.is_lvis:
-                    stats["metrics/APr(B)"] = val.stats_as_dict["APr"]
-                    stats["metrics/APc(B)"] = val.stats_as_dict["APc"]
-                    stats["metrics/APf(B)"] = val.stats_as_dict["APf"]
-                    stats["fitness"] = val.stats_as_dict["AP_all"]
+                    stats["fitness"] = stats["metrics/mAP50-95(B)"]  # always use box mAP50-95 for fitness
             except Exception as e:
                 LOGGER.warning(f"faster-coco-eval unable to run: {e}")
         return stats

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/models/yolo/model.py

@@ -406,18 +406,18 @@ class YOLOE(Model):
                 f"Expected equal number of bounding boxes and classes, but got {len(visual_prompts['bboxes'])} and "
                 f"{len(visual_prompts['cls'])} respectively"
             )
-        self.predictor = (predictor or self._smart_load("predictor"))(
-            overrides={
-                "task": self.model.task,
-                "mode": "predict",
-                "save": False,
-                "verbose": refer_image is None,
-                "batch": 1,
-            },
-            _callbacks=self.callbacks,
-        )
+            if not isinstance(self.predictor, yolo.yoloe.YOLOEVPDetectPredictor):
+                self.predictor = (predictor or yolo.yoloe.YOLOEVPDetectPredictor)(
+                    overrides={
+                        "task": self.model.task,
+                        "mode": "predict",
+                        "save": False,
+                        "verbose": refer_image is None,
+                        "batch": 1,
+                    },
+                    _callbacks=self.callbacks,
+                )
 
-        if len(visual_prompts):
             num_cls = (
                 max(len(set(c)) for c in visual_prompts["cls"])
                 if isinstance(source, list) and refer_image is None  # means multiple images
@@ -426,18 +426,19 @@ class YOLOE(Model):
             self.model.model[-1].nc = num_cls
             self.model.names = [f"object{i}" for i in range(num_cls)]
             self.predictor.set_prompts(visual_prompts.copy())
-
-        self.predictor.setup_model(model=self.model)
-
-        if refer_image is None and source is not None:
-            dataset = load_inference_source(source)
-            if dataset.mode in {"video", "stream"}:
-                # NOTE: set the first frame as refer image for videos/streams inference
-                refer_image = next(iter(dataset))[1][0]
-        if refer_image is not None and len(visual_prompts):
-            vpe = self.predictor.get_vpe(refer_image)
-            self.model.set_classes(self.model.names, vpe)
-            self.task = "segment" if isinstance(self.predictor, yolo.segment.SegmentationPredictor) else "detect"
-            self.predictor = None  # reset predictor
+            self.predictor.setup_model(model=self.model)
+
+            if refer_image is None and source is not None:
+                dataset = load_inference_source(source)
+                if dataset.mode in {"video", "stream"}:
+                    # NOTE: set the first frame as refer image for videos/streams inference
+                    refer_image = next(iter(dataset))[1][0]
+            if refer_image is not None:
+                vpe = self.predictor.get_vpe(refer_image)
+                self.model.set_classes(self.model.names, vpe)
+                self.task = "segment" if isinstance(self.predictor, yolo.segment.SegmentationPredictor) else "detect"
+                self.predictor = None  # reset predictor
+        elif isinstance(self.predictor, yolo.yoloe.YOLOEVPDetectPredictor):
+            self.predictor = None  # reset predictor if no visual prompts
 
         return super().predict(source, stream, **kwargs)

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/models/yolo/pose/val.py

@@ -8,7 +8,6 @@ import torch
 
 from ultralytics.models.yolo.detect import DetectionValidator
 from ultralytics.utils import LOGGER, ops
-from ultralytics.utils.checks import check_requirements
 from ultralytics.utils.metrics import OKS_SIGMA, PoseMetrics, kpt_iou
 
 
@@ -289,29 +288,6 @@ class PoseValidator(DetectionValidator):
 
     def eval_json(self, stats: Dict[str, Any]) -> Dict[str, Any]:
         """Evaluate object detection model using COCO JSON format."""
-        if self.args.save_json and self.is_coco and len(self.jdict):
-            anno_json = self.data["path"] / "annotations/person_keypoints_val2017.json"  # annotations
-            pred_json = self.save_dir / "predictions.json"  # predictions
-            LOGGER.info(f"\nEvaluating faster-coco-eval mAP using {pred_json} and {anno_json}...")
-            try:
-                check_requirements("faster-coco-eval>=1.6.7")
-                from faster_coco_eval import COCO, COCOeval_faster
-
-                for x in anno_json, pred_json:
-                    assert x.is_file(), f"{x} file not found"
-                anno = COCO(anno_json)  # init annotations api
-                pred = anno.loadRes(pred_json)  # init predictions api (must pass string, not Path)
-                kwargs = dict(cocoGt=anno, cocoDt=pred, print_function=LOGGER.info)
-                for i, eval in enumerate(
-                    [COCOeval_faster(iouType="bbox", **kwargs), COCOeval_faster(iouType="keypoints", **kwargs)]
-                ):
-                    eval.params.imgIds = [int(Path(x).stem) for x in self.dataloader.dataset.im_files]  # im to eval
-                    eval.evaluate()
-                    eval.accumulate()
-                    eval.summarize()
-                    idx = i * 4 + 2
-                    # update mAP50-95 and mAP50
-                    stats[self.metrics.keys[idx + 1]], stats[self.metrics.keys[idx]] = eval.stats[:2]
-            except Exception as e:
-                LOGGER.warning(f"faster-coco-eval unable to run: {e}")
-        return stats
+        anno_json = self.data["path"] / "annotations/person_keypoints_val2017.json"  # annotations
+        pred_json = self.save_dir / "predictions.json"  # predictions
+        return super().coco_evaluate(stats, pred_json, anno_json, ["bbox", "keypoints"], suffix=["Box", "Pose"])

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/models/yolo/segment/val.py

@@ -272,45 +272,10 @@ class SegmentationValidator(DetectionValidator):
 
     def eval_json(self, stats: Dict[str, Any]) -> Dict[str, Any]:
         """Return COCO-style instance segmentation evaluation metrics."""
-        if self.args.save_json and (self.is_lvis or self.is_coco) and len(self.jdict):
-            pred_json = self.save_dir / "predictions.json"  # predictions
-            anno_json = (
-                self.data["path"]
-                / "annotations"
-                / ("instances_val2017.json" if self.is_coco else f"lvis_v1_{self.args.split}.json")
-            )  # annotations
-
-            LOGGER.info(f"\nEvaluating faster-coco-eval mAP using {pred_json} and {anno_json}...")
-            try:
-                for x in anno_json, pred_json:
-                    assert x.is_file(), f"{x} file not found"
-                check_requirements("faster-coco-eval>=1.6.7")
-                from faster_coco_eval import COCO, COCOeval_faster
-
-                anno = COCO(anno_json)  # init annotations api
-                pred = anno.loadRes(pred_json)  # init predictions api (must pass string, not Path)
-                kwargs = dict(cocoGt=anno, cocoDt=pred, lvis_style=self.is_lvis, print_function=LOGGER.info)
-                for i, eval in enumerate(
-                    [COCOeval_faster(iouType="bbox", **kwargs), COCOeval_faster(iouType="segm", **kwargs)]
-                ):
-                    eval.params.imgIds = [int(Path(x).stem) for x in self.dataloader.dataset.im_files]  # im to eval
-                    eval.evaluate()
-                    eval.accumulate()
-                    eval.summarize()
-                    idx = i * 4 + 2
-                    # update mAP50-95 and mAP50
-                    stats[self.metrics.keys[idx + 1]] = eval.stats_as_dict["AP_all"]
-                    stats[self.metrics.keys[idx]] = eval.stats_as_dict["AP_50"]
-
-                    if self.is_lvis:
-                        tag = "B" if i == 0 else "M"
-                        stats[f"metrics/APr({tag})"] = eval.stats_as_dict["APr"]
-                        stats[f"metrics/APc({tag})"] = eval.stats_as_dict["APc"]
-                        stats[f"metrics/APf({tag})"] = eval.stats_as_dict["APf"]
-
-                if self.is_lvis:
-                    stats["fitness"] = stats["metrics/mAP50-95(B)"]
-
-            except Exception as e:
-                LOGGER.warning(f"faster-coco-eval unable to run: {e}")
-        return stats
+        pred_json = self.save_dir / "predictions.json"  # predictions
+        anno_json = (
+            self.data["path"]
+            / "annotations"
+            / ("instances_val2017.json" if self.is_coco else f"lvis_v1_{self.args.split}.json")
+        )  # annotations
+        return super().coco_evaluate(stats, pred_json, anno_json, ["bbox", "segm"], suffix=["Box", "Mask"])

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/models/yolo/world/train.py

@@ -158,7 +158,7 @@ class WorldTrainer(DetectionTrainer):
                 return txt_map
         LOGGER.info(f"Caching text embeddings to '{cache_path}'")
         assert self.model is not None
-        txt_feats = self.model.get_text_pe(texts, batch, cache_clip_model=False)
+        txt_feats = de_parallel(self.model).get_text_pe(texts, batch, cache_clip_model=False)
         txt_map = dict(zip(texts, txt_feats.squeeze(0)))
         torch.save(txt_map, cache_path)
         return txt_map

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/models/yolo/yoloe/train.py

@@ -222,7 +222,7 @@ class YOLOETrainerFromScratch(YOLOETrainer, WorldTrainerFromScratch):
                 return txt_map
         LOGGER.info(f"Caching text embeddings to '{cache_path}'")
         assert self.model is not None
-        txt_feats = self.model.get_text_pe(texts, batch, without_reprta=True, cache_clip_model=False)
+        txt_feats = de_parallel(self.model).get_text_pe(texts, batch, without_reprta=True, cache_clip_model=False)
         txt_map = dict(zip(texts, txt_feats.squeeze(0)))
         torch.save(txt_map, cache_path)
         return txt_map

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/nn/autobackend.py

@@ -259,7 +259,11 @@ class AutoBackend(nn.Module):
                 session = onnxruntime.InferenceSession(w, providers=providers)
             else:
                 check_requirements(
-                    ["model-compression-toolkit>=2.3.0", "sony-custom-layers[torch]>=0.3.0", "onnxruntime-extensions"]
+                    [
+                        "model-compression-toolkit>=2.3.0,<2.4.1",
+                        "sony-custom-layers[torch]>=0.3.0",
+                        "onnxruntime-extensions",
+                    ]
                 )
                 w = next(Path(w).glob("*.onnx"))
                 LOGGER.info(f"Loading {w} for ONNX IMX inference...")

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/nn/text_model.py

@@ -6,6 +6,7 @@ from typing import List, Union
 
 import torch
 import torch.nn as nn
+from PIL import Image
 
 from ultralytics.utils import checks
 from ultralytics.utils.torch_utils import smart_inference_mode
@@ -68,7 +69,7 @@ class CLIP(TextModel):
         >>> print(text_features.shape)
     """
 
-    def __init__(self, size: str, device: torch.device):
+    def __init__(self, size: str, device: torch.device) -> None:
         """
         Initialize the CLIP text encoder.
 
@@ -85,12 +86,12 @@ class CLIP(TextModel):
             >>> text_features = clip_model.encode_text(["a photo of a cat", "a photo of a dog"])
         """
         super().__init__()
-        self.model = clip.load(size, device=device)[0]
+        self.model, self.image_preprocess = clip.load(size, device=device)
         self.to(device)
         self.device = device
         self.eval()
 
-    def tokenize(self, texts: Union[str, List[str]]):
+    def tokenize(self, texts: Union[str, List[str]]) -> torch.Tensor:
         """
         Convert input texts to CLIP tokens.
 
@@ -108,7 +109,7 @@ class CLIP(TextModel):
         return clip.tokenize(texts).to(self.device)
 
     @smart_inference_mode()
-    def encode_text(self, texts: torch.Tensor, dtype: torch.dtype = torch.float32):
+    def encode_text(self, texts: torch.Tensor, dtype: torch.dtype = torch.float32) -> torch.Tensor:
         """
         Encode tokenized texts into normalized feature vectors.
 
@@ -133,6 +134,38 @@ class CLIP(TextModel):
         txt_feats = txt_feats / txt_feats.norm(p=2, dim=-1, keepdim=True)
         return txt_feats
 
+    @smart_inference_mode()
+    def encode_image(self, image: Union[Image.Image, torch.Tensor], dtype: torch.dtype = torch.float32) -> torch.Tensor:
+        """
+        Encode preprocessed images into normalized feature vectors.
+
+        This method processes preprocessed image inputs through the CLIP model to generate feature vectors, which are then
+        normalized to unit length. These normalized vectors can be used for text-image similarity comparisons.
+
+        Args:
+            image (PIL.Image | torch.Tensor): Preprocessed image input. If a PIL Image is provided, it will be
+                converted to a tensor using the model's image preprocessing function.
+            dtype (torch.dtype, optional): Data type for output features.
+
+        Returns:
+            (torch.Tensor): Normalized image feature vectors with unit length (L2 norm = 1).
+
+        Examples:
+            >>> from ultralytics.nn.text_model import CLIP
+            >>> from PIL import Image
+            >>> clip_model = CLIP("ViT-B/32", device="cuda")
+            >>> image = Image.open("path/to/image.jpg")
+            >>> image_tensor = clip_model.image_preprocess(image).unsqueeze(0).to("cuda")
+            >>> features = clip_model.encode_image(image_tensor)
+            >>> features.shape
+            torch.Size([1, 512])
+        """
+        if isinstance(image, Image.Image):
+            image = self.image_preprocess(image).unsqueeze(0).to(self.device)
+        img_feats = self.model.encode_image(image).to(dtype)
+        img_feats = img_feats / img_feats.norm(p=2, dim=-1, keepdim=True)
+        return img_feats
+
 
 class MobileCLIP(TextModel):
     """
@@ -160,7 +193,7 @@ class MobileCLIP(TextModel):
 
     config_size_map = {"s0": "s0", "s1": "s1", "s2": "s2", "b": "b", "blt": "b"}
 
-    def __init__(self, size: str, device: torch.device):
+    def __init__(self, size: str, device: torch.device) -> None:
         """
         Initialize the MobileCLIP text encoder.
 
@@ -201,7 +234,7 @@ class MobileCLIP(TextModel):
         self.device = device
         self.eval()
 
-    def tokenize(self, texts: List[str]):
+    def tokenize(self, texts: List[str]) -> torch.Tensor:
         """
         Convert input texts to MobileCLIP tokens.
 
@@ -218,7 +251,7 @@ class MobileCLIP(TextModel):
         return self.tokenizer(texts).to(self.device)
 
     @smart_inference_mode()
-    def encode_text(self, texts: torch.Tensor, dtype: torch.dtype = torch.float32):
+    def encode_text(self, texts: torch.Tensor, dtype: torch.dtype = torch.float32) -> torch.Tensor:
         """
         Encode tokenized texts into normalized feature vectors.
 
@@ -286,7 +319,7 @@ class MobileCLIPTS(TextModel):
         self.tokenizer = clip.clip.tokenize
         self.device = device
 
-    def tokenize(self, texts: List[str]):
+    def tokenize(self, texts: List[str]) -> torch.Tensor:
         """
         Convert input texts to MobileCLIP tokens.
 
@@ -303,7 +336,7 @@ class MobileCLIPTS(TextModel):
         return self.tokenizer(texts).to(self.device)
 
     @smart_inference_mode()
-    def encode_text(self, texts: torch.Tensor, dtype: torch.dtype = torch.float32):
+    def encode_text(self, texts: torch.Tensor, dtype: torch.dtype = torch.float32) -> torch.Tensor:
         """
         Encode tokenized texts into normalized feature vectors.
 
@@ -322,10 +355,10 @@ class MobileCLIPTS(TextModel):
             torch.Size([2, 512])  # Actual dimension depends on model size
         """
         # NOTE: no need to do normalization here as it's embedded in the torchscript model
-        return self.encoder(texts)
+        return self.encoder(texts).to(dtype)
 
 
-def build_text_model(variant: str, device: torch.device = None):
+def build_text_model(variant: str, device: torch.device = None) -> TextModel:
     """
     Build a text encoding model based on the specified variant.
 

{ultralytics-8.3.158 → ultralytics-8.3.160}/ultralytics/solutions/heatmap.py

@@ -124,6 +124,6 @@ class Heatmap(ObjectCounter):
             plot_im=plot_im,
             in_count=self.in_count,
             out_count=self.out_count,
-            classwise_count=dict(self.classwise_counts),
+            classwise_count=dict(self.classwise_count),
             total_tracks=len(self.track_ids),
         )