ultralytics 8.0.227__tar.gz → 8.0.229__tar.gz

{ultralytics-8.0.227/ultralytics.egg-info → ultralytics-8.0.229}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ultralytics
-Version: 8.0.227
+Version: 8.0.229
 Summary: Ultralytics YOLOv8 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification.
 Home-page: https://github.com/ultralytics/ultralytics
 Author: Ultralytics
@@ -60,6 +60,8 @@ Provides-Extra: export
 Requires-Dist: coremltools>=7.0; extra == "export"
 Requires-Dist: openvino-dev>=2023.0; extra == "export"
 Requires-Dist: tensorflow<=2.13.1; extra == "export"
+Requires-Dist: jax<=0.4.21; extra == "export"
+Requires-Dist: jaxlib<=0.4.21; extra == "export"
 Requires-Dist: tensorflowjs; extra == "export"
 
 <div align="center">

{ultralytics-8.0.227 → ultralytics-8.0.229}/requirements.txt

@@ -32,6 +32,8 @@ seaborn>=0.11.0
 # scikit-learn==0.19.2  # CoreML quantization
 # tensorflow>=2.4.1,<=2.13.1  # TF exports (-cpu, -aarch64, -macos)
 # tflite-support
+# jax<=0.4.21  # tensorflowjs bug https://github.com/google/jax/issues/18978
+# jaxlib<=0.4.21  # tensorflowjs bug https://github.com/google/jax/issues/18978
 # tensorflowjs>=3.9.0  # TF.js export
 # openvino-dev>=2023.0  # OpenVINO export
 

{ultralytics-8.0.227 → ultralytics-8.0.229}/setup.py

@@ -81,6 +81,8 @@ setup(
             'coremltools>=7.0',
             'openvino-dev>=2023.0',
             'tensorflow<=2.13.1',  # TF bug https://github.com/ultralytics/ultralytics/issues/5161
+            'jax<=0.4.21',  # tensorflowjs bug https://github.com/google/jax/issues/18978
+            'jaxlib<=0.4.21',  # tensorflowjs bug https://github.com/google/jax/issues/18978
             'tensorflowjs',  # automatically installs tensorflow
         ], },
     classifiers=[

{ultralytics-8.0.227 → ultralytics-8.0.229}/tests/test_cuda.py

@@ -61,6 +61,7 @@ def test_autobatch():
     check_train_batch_size(YOLO(MODEL).model.cuda(), imgsz=128, amp=True)
 
 
+@pytest.mark.slow
 @pytest.mark.skipif(not CUDA_IS_AVAILABLE, reason='CUDA is not available')
 def test_utils_benchmarks():
     """Profile YOLO models for performance benchmarks."""

{ultralytics-8.0.227 → ultralytics-8.0.229}/tests/test_python.py

@@ -511,3 +511,13 @@ def test_model_tune():
     """Tune YOLO model for performance."""
     YOLO('yolov8n-pose.pt').tune(data='coco8-pose.yaml', plots=False, imgsz=32, epochs=1, iterations=2, device='cpu')
     YOLO('yolov8n-cls.pt').tune(data='imagenet10', plots=False, imgsz=32, epochs=1, iterations=2, device='cpu')
+
+
+def test_model_embeddings():
+    """Test YOLO model embeddings."""
+    model_detect = YOLO(MODEL)
+    model_segment = YOLO(WEIGHTS_DIR / 'yolov8n-seg.pt')
+
+    for batch in [SOURCE], [SOURCE, SOURCE]:  # test batch size 1 and 2
+        assert len(model_detect.embed(source=batch, imgsz=32)) == len(batch)
+        assert len(model_segment.embed(source=batch, imgsz=32)) == len(batch)

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/__init__.py

@@ -1,6 +1,6 @@
 # Ultralytics YOLO 🚀, AGPL-3.0 license
 
-__version__ = '8.0.227'
+__version__ = '8.0.229'
 
 from ultralytics.models import RTDETR, SAM, YOLO
 from ultralytics.models.fastsam import FastSAM

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/cfg/__init__.py

@@ -63,7 +63,7 @@ CLI_HELP_MSG = \
     """
 
 # Define keys for arg type checks
-CFG_FLOAT_KEYS = 'warmup_epochs', 'box', 'cls', 'dfl', 'degrees', 'shear'
+CFG_FLOAT_KEYS = 'warmup_epochs', 'box', 'cls', 'dfl', 'degrees', 'shear', 'time'
 CFG_FRACTION_KEYS = ('dropout', 'iou', 'lr0', 'lrf', 'momentum', 'weight_decay', 'warmup_momentum', 'warmup_bias_lr',
                      'label_smoothing', 'hsv_h', 'hsv_s', 'hsv_v', 'translate', 'scale', 'perspective', 'flipud',
                      'fliplr', 'mosaic', 'mixup', 'copy_paste', 'conf', 'iou', 'fraction')  # fraction floats 0.0 - 1.0

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/cfg/default.yaml

@@ -8,6 +8,7 @@ mode: train  # (str) YOLO mode, i.e. train, val, predict, export, track, benchma
 model:  # (str, optional) path to model file, i.e. yolov8n.pt, yolov8n.yaml
 data:  # (str, optional) path to data file, i.e. coco128.yaml
 epochs: 100  # (int) number of epochs to train for
+time:  # (float, optional) number of hours to train for, overrides epochs if supplied
 patience: 50  # (int) epochs to wait for no observable improvement for early stopping of training
 batch: 16  # (int) number of images per batch (-1 for AutoBatch)
 imgsz: 640  # (int | list) input images size as int for train and val modes, or list[w,h] for predict and export modes
@@ -60,6 +61,7 @@ augment: False  # (bool) apply image augmentation to prediction sources
 agnostic_nms: False  # (bool) class-agnostic NMS
 classes:  # (int | list[int], optional) filter results by class, i.e. classes=0, or classes=[0,2,3]
 retina_masks: False  # (bool) use high-resolution segmentation masks
+embed:  # (list[int], optional) return feature vectors/embeddings from given layers
 
 # Visualize settings ---------------------------------------------------------------------------------------------------
 show: False  # (bool) show predicted images and videos if environment allows

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/data/build.py

@@ -100,7 +100,7 @@ def build_dataloader(dataset, batch, workers, shuffle=True, rank=-1):
     """Return an InfiniteDataLoader or DataLoader for training or validation set."""
     batch = min(batch, len(dataset))
     nd = torch.cuda.device_count()  # number of CUDA devices
-    nw = min([os.cpu_count() // max(nd, 1), batch if batch > 1 else 0, workers])  # number of workers
+    nw = min([os.cpu_count() // max(nd, 1), batch, workers])  # number of workers
     sampler = None if rank == -1 else distributed.DistributedSampler(dataset, shuffle=shuffle)
     generator = torch.Generator()
     generator.manual_seed(6148914691236517205 + RANK)

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/engine/exporter.py

@@ -459,11 +459,14 @@ class Exporter:
                 f'{prefix} WARNING ⚠️ 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.')
-            _, assets = get_github_assets(repo='pnnx/pnnx', retry=True)
-            system = 'macos' if MACOS else 'ubuntu' if LINUX else 'windows'  # operating system
-            asset = [x for x in assets if system in x][0] if assets else \
-                f'https://github.com/pnnx/pnnx/releases/download/20230816/pnnx-20230816-{system}.zip'  # fallback
-            asset = attempt_download_asset(asset, repo='pnnx/pnnx', release='latest')
+            system = ['macos'] if MACOS else ['windows'] if WINDOWS else ['ubuntu', 'linux']  # operating system
+            try:
+                _, assets = get_github_assets(repo='pnnx/pnnx', retry=True)
+                url = [x for x in assets if any(s in x for s in system)][0]
+            except Exception as e:
+                url = f'https://github.com/pnnx/pnnx/releases/download/20231127/pnnx-20231127-{system[0]}.zip'
+                LOGGER.warning(f'{prefix} WARNING ⚠️ PNNX GitHub assets not found: {e}, using default {url}')
+            asset = attempt_download_asset(url, repo='pnnx/pnnx', release='latest')
             if check_is_path_safe(Path.cwd(), asset):  # avoid path traversal security vulnerability
                 unzip_dir = Path(asset).with_suffix('')
                 (unzip_dir / name).rename(pnnx)  # move binary to ROOT
@@ -781,7 +784,8 @@ class Exporter:
     @try_export
     def export_tfjs(self, prefix=colorstr('TensorFlow.js:')):
         """YOLOv8 TensorFlow.js export."""
-        check_requirements('tensorflowjs')
+        # JAX bug requiring install constraints in https://github.com/google/jax/issues/18978
+        check_requirements(['jax<=0.4.21', 'jaxlib<=0.4.21', 'tensorflowjs'])
         import tensorflow as tf
         import tensorflowjs as tfjs  # noqa
 
@@ -795,8 +799,9 @@ class Exporter:
         outputs = ','.join(gd_outputs(gd))
         LOGGER.info(f'\n{prefix} output node names: {outputs}')
 
+        quantization = '--quantize_float16' if self.args.half else '--quantize_uint8' if self.args.int8 else ''
         with spaces_in_path(f_pb) as fpb_, spaces_in_path(f) as f_:  # exporter can not handle spaces in path
-            cmd = f'tensorflowjs_converter --input_format=tf_frozen_model --output_node_names={outputs} "{fpb_}" "{f_}"'
+            cmd = f'tensorflowjs_converter --input_format=tf_frozen_model {quantization} --output_node_names={outputs} "{fpb_}" "{f_}"'
             LOGGER.info(f"{prefix} running '{cmd}'")
             subprocess.run(cmd, shell=True)
 

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/engine/model.py

@@ -94,7 +94,7 @@ class Model(nn.Module):
             self._load(model, task)
 
     def __call__(self, source=None, stream=False, **kwargs):
-        """Calls the 'predict' function with given arguments to perform object detection."""
+        """Calls the predict() method with given arguments to perform object detection."""
         return self.predict(source, stream, **kwargs)
 
     @staticmethod
@@ -201,6 +201,24 @@ class Model(nn.Module):
         self._check_is_pytorch_model()
         self.model.fuse()
 
+    def embed(self, source=None, stream=False, **kwargs):
+        """
+        Calls the predict() method and returns image embeddings.
+
+        Args:
+            source (str | int | PIL | np.ndarray): The source of the image to make predictions on.
+                Accepts all source types accepted by the YOLO model.
+            stream (bool): Whether to stream the predictions or not. Defaults to False.
+            **kwargs : Additional keyword arguments passed to the predictor.
+                Check the 'configuration' section in the documentation for all available options.
+
+        Returns:
+            (List[torch.Tensor]): A list of image embeddings.
+        """
+        if not kwargs.get('embed'):
+            kwargs['embed'] = [len(self.model.model) - 2]  # embed second-to-last layer if no indices passed
+        return self.predict(source, stream, **kwargs)
+
     def predict(self, source=None, stream=False, predictor=None, **kwargs):
         """
         Perform prediction using the YOLO model.

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/engine/predictor.py

@@ -134,7 +134,7 @@ class BasePredictor:
         """Runs inference on a given image using the specified model and arguments."""
         visualize = increment_path(self.save_dir / Path(self.batch[0][0]).stem,
                                    mkdir=True) if self.args.visualize and (not self.source_type.tensor) else False
-        return self.model(im, augment=self.args.augment, visualize=visualize)
+        return self.model(im, augment=self.args.augment, visualize=visualize, embed=self.args.embed, *args, **kwargs)
 
     def pre_transform(self, im):
         """
@@ -263,6 +263,9 @@ class BasePredictor:
                 # Inference
                 with profilers[1]:
                     preds = self.inference(im, *args, **kwargs)
+                    if self.args.embed:
+                        yield from [preds] if isinstance(preds, torch.Tensor) else preds  # yield embedding tensors
+                        continue
 
                 # Postprocess
                 with profilers[2]:

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/engine/trainer.py

@@ -189,6 +189,14 @@ class BaseTrainer:
         else:
             self._do_train(world_size)
 
+    def _setup_scheduler(self):
+        """Initialize training learning rate scheduler."""
+        if self.args.cos_lr:
+            self.lf = one_cycle(1, self.args.lrf, self.epochs)  # cosine 1->hyp['lrf']
+        else:
+            self.lf = lambda x: max(1 - x / self.epochs, 0) * (1.0 - self.args.lrf) + self.args.lrf  # linear
+        self.scheduler = optim.lr_scheduler.LambdaLR(self.optimizer, lr_lambda=self.lf)
+
     def _setup_ddp(self, world_size):
         """Initializes and sets the DistributedDataParallel parameters for training."""
         torch.cuda.set_device(RANK)
@@ -269,11 +277,7 @@ class BaseTrainer:
                                               decay=weight_decay,
                                               iterations=iterations)
         # Scheduler
-        if self.args.cos_lr:
-            self.lf = one_cycle(1, self.args.lrf, self.epochs)  # cosine 1->hyp['lrf']
-        else:
-            self.lf = lambda x: (1 - x / self.epochs) * (1.0 - self.args.lrf) + self.args.lrf  # linear
-        self.scheduler = optim.lr_scheduler.LambdaLR(self.optimizer, lr_lambda=self.lf)
+        self._setup_scheduler()
         self.stopper, self.stop = EarlyStopping(patience=self.args.patience), False
         self.resume_training(ckpt)
         self.scheduler.last_epoch = self.start_epoch - 1  # do not move
@@ -285,17 +289,18 @@ class BaseTrainer:
             self._setup_ddp(world_size)
         self._setup_train(world_size)
 
-        self.epoch_time = None
-        self.epoch_time_start = time.time()
-        self.train_time_start = time.time()
         nb = len(self.train_loader)  # number of batches
         nw = max(round(self.args.warmup_epochs * nb), 100) if self.args.warmup_epochs > 0 else -1  # warmup iterations
         last_opt_step = -1
+        self.epoch_time = None
+        self.epoch_time_start = time.time()
+        self.train_time_start = time.time()
         self.run_callbacks('on_train_start')
         LOGGER.info(f'Image sizes {self.args.imgsz} train, {self.args.imgsz} val\n'
                     f'Using {self.train_loader.num_workers * (world_size or 1)} dataloader workers\n'
                     f"Logging results to {colorstr('bold', self.save_dir)}\n"
-                    f'Starting training for {self.epochs} epochs...')
+                    f'Starting training for '
+                    f'{self.args.time} hours...' if self.args.time else f'{self.epochs} epochs...')
         if self.args.close_mosaic:
             base_idx = (self.epochs - self.args.close_mosaic) * nb
             self.plot_idx.extend([base_idx, base_idx + 1, base_idx + 2])
@@ -323,7 +328,7 @@ class BaseTrainer:
                 ni = i + nb * epoch
                 if ni <= nw:
                     xi = [0, nw]  # x interp
-                    self.accumulate = max(1, np.interp(ni, xi, [1, self.args.nbs / self.batch_size]).round())
+                    self.accumulate = max(1, int(np.interp(ni, xi, [1, self.args.nbs / self.batch_size]).round()))
                     for j, x in enumerate(self.optimizer.param_groups):
                         # Bias lr falls from 0.1 to lr0, all other lrs rise from 0.0 to lr0
                         x['lr'] = np.interp(
@@ -348,6 +353,16 @@ class BaseTrainer:
                     self.optimizer_step()
                     last_opt_step = ni
 
+                    # Timed stopping
+                    if self.args.time:
+                        self.stop = (time.time() - self.train_time_start) > (self.args.time * 3600)
+                        if RANK != -1:  # if DDP training
+                            broadcast_list = [self.stop if RANK == 0 else None]
+                            dist.broadcast_object_list(broadcast_list, 0)  # broadcast 'stop' to all ranks
+                            self.stop = broadcast_list[0]
+                        if self.stop:  # training time exceeded
+                            break
+
                 # Log
                 mem = f'{torch.cuda.memory_reserved() / 1E9 if torch.cuda.is_available() else 0:.3g}G'  # (GB)
                 loss_len = self.tloss.shape[0] if len(self.tloss.size()) else 1
@@ -363,31 +378,37 @@ class BaseTrainer:
                 self.run_callbacks('on_train_batch_end')
 
             self.lr = {f'lr/pg{ir}': x['lr'] for ir, x in enumerate(self.optimizer.param_groups)}  # for loggers
-
-            with warnings.catch_warnings():
-                warnings.simplefilter('ignore')  # suppress 'Detected lr_scheduler.step() before optimizer.step()'
-                self.scheduler.step()
             self.run_callbacks('on_train_epoch_end')
-
             if RANK in (-1, 0):
-
-                # Validation
+                final_epoch = epoch + 1 == self.epochs
                 self.ema.update_attr(self.model, include=['yaml', 'nc', 'args', 'names', 'stride', 'class_weights'])
-                final_epoch = (epoch + 1 == self.epochs) or self.stopper.possible_stop
 
-                if self.args.val or final_epoch:
+                # Validation
+                if self.args.val or final_epoch or self.stopper.possible_stop or self.stop:
                     self.metrics, self.fitness = self.validate()
                 self.save_metrics(metrics={**self.label_loss_items(self.tloss), **self.metrics, **self.lr})
-                self.stop = self.stopper(epoch + 1, self.fitness)
+                self.stop |= self.stopper(epoch + 1, self.fitness)
+                if self.args.time:
+                    self.stop |= (time.time() - self.train_time_start) > (self.args.time * 3600)
 
                 # Save model
-                if self.args.save or (epoch + 1 == self.epochs):
+                if self.args.save or final_epoch:
                     self.save_model()
                     self.run_callbacks('on_model_save')
 
-            tnow = time.time()
-            self.epoch_time = tnow - self.epoch_time_start
-            self.epoch_time_start = tnow
+            # Scheduler
+            t = time.time()
+            self.epoch_time = t - self.epoch_time_start
+            self.epoch_time_start = t
+            with warnings.catch_warnings():
+                warnings.simplefilter('ignore')  # suppress 'Detected lr_scheduler.step() before optimizer.step()'
+                if self.args.time:
+                    mean_epoch_time = (t - self.train_time_start) / (epoch - self.start_epoch + 1)
+                    self.epochs = self.args.epochs = math.ceil(self.args.time * 3600 / mean_epoch_time)
+                    self._setup_scheduler()
+                    self.scheduler.last_epoch = self.epoch  # do not move
+                    self.stop |= epoch >= self.epochs  # stop if exceeded epochs
+                self.scheduler.step()
             self.run_callbacks('on_fit_epoch_end')
             torch.cuda.empty_cache()  # clear GPU memory at end of epoch, may help reduce CUDA out of memory errors
 
@@ -395,8 +416,7 @@ class BaseTrainer:
             if RANK != -1:  # if DDP training
                 broadcast_list = [self.stop if RANK == 0 else None]
                 dist.broadcast_object_list(broadcast_list, 0)  # broadcast 'stop' to all ranks
-                if RANK != 0:
-                    self.stop = broadcast_list[0]
+                self.stop = broadcast_list[0]
             if self.stop:
                 break  # must break all DDP ranks
 

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/nn/autobackend.py

@@ -333,7 +333,7 @@ class AutoBackend(nn.Module):
 
         self.__dict__.update(locals())  # assign all variables to self
 
-    def forward(self, im, augment=False, visualize=False):
+    def forward(self, im, augment=False, visualize=False, embed=None):
         """
         Runs inference on the YOLOv8 MultiBackend model.
 
@@ -341,6 +341,7 @@ class AutoBackend(nn.Module):
             im (torch.Tensor): The image tensor to perform inference on.
             augment (bool): whether to perform data augmentation during inference, defaults to False
             visualize (bool): whether to visualize the output predictions, defaults to False
+            embed (list, optional): A list of feature vectors/embeddings to return.
 
         Returns:
             (tuple): Tuple containing the raw output tensor, and processed output for visualization (if visualize=True)
@@ -352,7 +353,7 @@ class AutoBackend(nn.Module):
             im = im.permute(0, 2, 3, 1)  # torch BCHW to numpy BHWC shape(1,320,192,3)
 
         if self.pt or self.nn_module:  # PyTorch
-            y = self.model(im, augment=augment, visualize=visualize) if augment or visualize else self.model(im)
+            y = self.model(im, augment=augment, visualize=visualize, embed=embed)
         elif self.jit:  # TorchScript
             y = self.model(im)
         elif self.dnn:  # ONNX OpenCV DNN

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/nn/tasks.py

@@ -41,7 +41,7 @@ class BaseModel(nn.Module):
             return self.loss(x, *args, **kwargs)
         return self.predict(x, *args, **kwargs)
 
-    def predict(self, x, profile=False, visualize=False, augment=False):
+    def predict(self, x, profile=False, visualize=False, augment=False, embed=None):
         """
         Perform a forward pass through the network.
 
@@ -50,15 +50,16 @@ class BaseModel(nn.Module):
             profile (bool):  Print the computation time of each layer if True, defaults to False.
             visualize (bool): Save the feature maps of the model if True, defaults to False.
             augment (bool): Augment image during prediction, defaults to False.
+            embed (list, optional): A list of feature vectors/embeddings to return.
 
         Returns:
             (torch.Tensor): The last output of the model.
         """
         if augment:
             return self._predict_augment(x)
-        return self._predict_once(x, profile, visualize)
+        return self._predict_once(x, profile, visualize, embed)
 
-    def _predict_once(self, x, profile=False, visualize=False):
+    def _predict_once(self, x, profile=False, visualize=False, embed=None):
         """
         Perform a forward pass through the network.
 
@@ -66,11 +67,12 @@ class BaseModel(nn.Module):
             x (torch.Tensor): The input tensor to the model.
             profile (bool):  Print the computation time of each layer if True, defaults to False.
             visualize (bool): Save the feature maps of the model if True, defaults to False.
+            embed (list, optional): A list of feature vectors/embeddings to return.
 
         Returns:
             (torch.Tensor): The last output of the model.
         """
-        y, dt = [], []  # outputs
+        y, dt, embeddings = [], [], []  # outputs
         for m in self.model:
             if m.f != -1:  # if not from previous layer
                 x = y[m.f] if isinstance(m.f, int) else [x if j == -1 else y[j] for j in m.f]  # from earlier layers
@@ -80,6 +82,10 @@ class BaseModel(nn.Module):
             y.append(x if m.i in self.save else None)  # save output
             if visualize:
                 feature_visualization(x, m.type, m.i, save_dir=visualize)
+            if embed and m.i in embed:
+                embeddings.append(nn.functional.adaptive_avg_pool2d(x, (1, 1)).squeeze(-1).squeeze(-1))  # flatten
+                if m.i == max(embed):
+                    return torch.unbind(torch.cat(embeddings, 1), dim=0)
         return x
 
     def _predict_augment(self, x):
@@ -454,7 +460,7 @@ class RTDETRDetectionModel(DetectionModel):
         return sum(loss.values()), torch.as_tensor([loss[k].detach() for k in ['loss_giou', 'loss_class', 'loss_bbox']],
                                                    device=img.device)
 
-    def predict(self, x, profile=False, visualize=False, batch=None, augment=False):
+    def predict(self, x, profile=False, visualize=False, batch=None, augment=False, embed=None):
         """
         Perform a forward pass through the model.
 
@@ -464,11 +470,12 @@ class RTDETRDetectionModel(DetectionModel):
             visualize (bool, optional): If True, save feature maps for visualization. Defaults to False.
             batch (dict, optional): Ground truth data for evaluation. Defaults to None.
             augment (bool, optional): If True, perform data augmentation during inference. Defaults to False.
+            embed (list, optional): A list of feature vectors/embeddings to return.
 
         Returns:
             (torch.Tensor): Model's output tensor.
         """
-        y, dt = [], []  # outputs
+        y, dt, embeddings = [], [], []  # outputs
         for m in self.model[:-1]:  # except the head part
             if m.f != -1:  # if not from previous layer
                 x = y[m.f] if isinstance(m.f, int) else [x if j == -1 else y[j] for j in m.f]  # from earlier layers
@@ -478,6 +485,10 @@ class RTDETRDetectionModel(DetectionModel):
             y.append(x if m.i in self.save else None)  # save output
             if visualize:
                 feature_visualization(x, m.type, m.i, save_dir=visualize)
+            if embed and m.i in embed:
+                embeddings.append(nn.functional.adaptive_avg_pool2d(x, (1, 1)).squeeze(-1).squeeze(-1))  # flatten
+                if m.i == max(embed):
+                    return torch.unbind(torch.cat(embeddings, 1), dim=0)
         head = self.model[-1]
         x = head([y[j] for j in head.f], batch)  # head inference
         return x

{ultralytics-8.0.227 → ultralytics-8.0.229}/ultralytics/solutions/ai_gym.py

@@ -62,7 +62,7 @@ class AIGym:
 
     def start_counting(self, im0, results, frame_count):
         """
-        function used to count the gym steps
+        Function used to count the gym steps
         Args:
             im0 (ndarray): Current frame from the video stream.
             results: Pose estimation data