Standard convolution with args(ch_in, ch_out, kernel, stride, padding, groups, dilation, activation).

YOLO (You Only Look Once) object detection model.

A class for storing and manipulating inference results. Args: orig_img (numpy.ndarray): The original image as a numpy array.

Resize image and padding for detection, instance segmentation, pose.

Standard bottleneck.

Class for computing evaluation metrics for YOLOv8 model. Attributes: p (list): Precision for each class. Shape: (nc,).

YOLOv8 Profile class. Usage: as a decorator with @Profile() or as a context manager with 'with Profile():'

An attention layer that allows for downscaling the size of the embedding after projection to queries, keys, and values.

Now only numpy is supported.

Dataset class for loading object detection and/or segmentation labels in YOLO format. Args: data (dict, optional): A dataset YAML di

Depth-wise convolution.

Ultralytics IterableSimpleNamespace is an extension class of SimpleNamespace that adds iterable functionality and enables usage with dict() a

Very simple multi-layer perceptron (also called FFN)

Ultralytics Annotator for train/val mosaics and JPGs and predictions annotations. Attributes: im (Image.Image or numpy array): The image

A class for storing and manipulating detection boxes. Args: boxes (torch.Tensor | numpy.ndarray): A tensor or numpy array containing

YOLOv8 CenterCrop class for image preprocessing, i.e. T.Compose([CenterCrop(size), ToTensor()])

YOLO Classification Dataset. Args: root (str): Dataset path. Attributes: cache_ram (bool): True if images should be cac

A class for calculating and updating a confusion matrix for object detection and classification tasks. Attributes: task (str): The t

Ghost Convolution https://github.com/huawei-noah/ghostnet.

For bytetrack A simple Kalman filter for tracking bounding boxes in image space. The 8-dimensional state space x, y, a, h, vx,

For BoT-SORT A simple Kalman filter for tracking bounding boxes in image space. The 8-dimensional state space x, y, w, h, vx, v

Lightly adapted from https://github.com/facebookresearch/MaskFormer/blob/main/mask_former/modeling/transformer/transformer_predictor.py

A class for storing and manipulating detection masks. Attributes: segments (list): Deprecated property for segments (normalized).

ProfileModels class for profiling different models on ONNX and TensorRT. This class profiles the performance of different models, provided t

SAM model interface.

Calculates and aggregates detection and segmentation metrics over a given set of classes. Args: save_dir (Path): Path to the directo

YOLOv8 ToTensor class for image preprocessing, i.e. T.Compose([LetterBox(size), ToTensor()]).

YOLOv8 Albumentations class (optional, only used if package is installed)

Multi-head Attention block with relative position embeddings.

A basic TinyViT layer for one stage. Args: dim (int): Number of input channels. input_resolution (tuple[int]): Input resolution.

Transformer blocks with support of window attention and residual propagation blocks

Channel-attention module https://github.com/open-mmlab/mmdetection/tree/v3.0.0rc1/configs/rtmdet.

YOLOv8 classification model.

YOLOv8 classification head, i.e. x(b,c1,20,20) to x(b,c2).

Class for computing classification metrics including top-1 and top-5 accuracy. Attributes: top1 (float): The top-1 accuracy.

Ultralytics default color palette https://ultralytics.com/. This class provides methods to work with the Ultralytics color palette, including con

Integral module of Distribution Focal Loss (DFL). Proposed in Generalized Focal Loss https://ieeexplore.ieee.org/document/9792391

https://github.com/PaddlePaddle/PaddleDetection/blob/develop/ppdet/modeling/transformers/deformable_transformer.py

https://github.com/PaddlePaddle/PaddleDetection/blob/develop/ppdet/modeling/transformers/deformable_transformer.py https://github.com/fundame

This class is a utility class for computing detection metrics such as precision, recall, and mean average precision (mAP) of an object detect

YOLOv8 detection model.

Early stopping class that stops training when a specified number of epochs have passed without improvement.

A custom logging filter class for removing emojis in log messages. This filter is particularly useful for ensuring compatibility with Window

Ensemble of models.

A class for collecting anonymous event analytics. Event analytics are enabled when sync=True in settings and disabled when sync=False. Run 'y

A class for exporting a model. Attributes: args (SimpleNamespace): Configuration for the exporter. save_dir (Path): Director

FastSAM model interface. Usage - Predict: from ultralytics import FastSAM model = FastSAM('last.pt') results = mode

Wraps focal loss around existing loss_fcn(), i.e. criteria = FocalLoss(nn.BCEWithLogitsLoss(), gamma=1.5).

Ghost Bottleneck https://github.com/huawei-noah/ghostnet.

A class for generating HUB dataset JSON and `-hub` dataset directory. Args: path (str): Path to data.yaml or data.zip (with data.yam

HUB training session for Ultralytics HUB YOLO models. Handles model initialization, heartbeats, and checkpointing. Args: url (str):

A module implementing the HungarianMatcher, which is a differentiable module to solve the assignment problem in an end-to-end fashion. H

Dataloader that reuses workers. Uses same syntax as vanilla DataLoader.

A class for storing and manipulating detection keypoints. Attributes: xy (torch.Tensor): A collection of keypoints containing x, y c

YOLOv8 image/video dataloader, i.e. `yolo predict source=image.jpg/vid.mp4`.

YOLOv8 screenshot dataloader, i.e. `yolo predict source=screen`.

YOLOv8 streamloader, i.e. `yolo predict source='rtsp://example.com/media.mp4' # RTSP, RTMP, HTTP streams`.

Original Multi-Scale Deformable Attention Module. https://github.com/fundamentalvision/Deformable-DETR/blob/main/models/ops/modules/ms_deform

Updated Exponential Moving Average (EMA) from https://github.com/rwightman/pytorch-image-models Keeps a moving average of everything in the model

Mosaic augmentation. This class performs mosaic augmentation by combining multiple (4 or 9) images into a single mosaic image. The augme