A conv block that bundles conv/norm/activation layers. This block simplifies the usage of convolution layers, which are commonly used with a

A wrapper around :class:`torch.nn.Conv2d` to support empty inputs and more features.

A registry to map strings to classes. Registered object could be built from registry. Example: >>> MODELS = Registry('models')

This class represents a list of instances in an image. It stores the attributes of instances (e.g., boxes, masks, labels, scores) as "fields"

This structure stores a list of boxes as a Nx4 torch.Tensor. It supports some common methods about boxes (`area`, `clip`, `nonempty`, etc

Compose multiple transforms sequentially. Args: transforms (Sequence[dict | callable]): Sequence of transform object or confi

ResLayer to build ResNet style backbone. Args: block (nn.Module): Residual block used to build ResLayer. num_blocks (int): Number

COCO pose estimation task evaluation metric. Evaluate AR, AP, and mAP for keypoint detection tasks. Support COCO dataset and other datasets i

Wrapper metric for evaluating pose metric on user-defined body parts. Sometimes one may be interested in the performance of a pose model on c

Top-down head introduced in RTMPose (2023). The head is composed of a large-kernel convolutional layer, a fully-connected layer and a Gated At

A simple structure that contains basic shape specification about a tensor. It is often used as the auxiliary inputs/outputs of models, to

Top-down heatmap head introduced in `Simple Baselines`_ by Xiao et al (2018). The head is composed of a few deconvolutional layers followed by a

ResNet backbone. Please refer to the `paper <https://arxiv.org/abs/1512.03385>`__ for details. Args: depth (int): Network depth,

ShuffleNetV1 backbone. Args: groups (int, optional): The number of groups to be used in grouped 1x1 convolutions in each Shuf

ViPNAS heatmap head introduced in `ViPNAS`_ by Xu et al (2021). The head is composed of a few deconvolutional layers followed by a convolutional

This class stores the segmentation masks for all objects in one image, in the form of bitmaps. Attributes: tensor: bool Tensor o

Multi-stage heatmap head introduced in `Convolutional Pose Machines`_ by Wei et al (2016) and used by `Stacked Hourglass Networks`_ by Newell et a

MobileNetV2 backbone. Args: widen_factor (float): Width multiplier, multiply number of channels in each layer by this amount.

Normlize image by given mean and std.

Randomly flip the image, bbox and keypoints. Required Keys: - img - img_shape - flip_indices - input_size (optio

ShuffleNetV2 backbone. Args: widen_factor (float): Width multiplier - adjusts the number of channels in each layer by this am

MobileNetV3 backbone. Args: arch (str): Architecture of mobilnetv3, from {small, big}. Default: small. conv_cfg (dict

This structure stores a list of rotated boxes as a Nx5 torch.Tensor. It supports some common methods about boxes (`area`, `clip`, `nonemp

SEResNet backbone. Please refer to the `paper <https://arxiv.org/abs/1709.01507>`__ for details. Args: depth (int): Network dept

PoseTrack18 evaluation metric. Evaluate AP, and mAP for keypoint detection tasks. Support PoseTrack18 (video) dataset. Please refer to `<

VGG backbone. Args: depth (int): Depth of vgg, from {11, 13, 16, 19}. with_norm (bool): Use BatchNorm or not. num_classes

ViPNAS_ResNet backbone. "ViPNAS: Efficient Video Pose Estimation via Neural Architecture Search" More details can be found in the `paper

COCO-WholeBody evaluation metric. Evaluate AR, AP, and mAP for COCO-WholeBody keypoint detection tasks. Support COCO-WholeBody dataset. Pleas

Top-down integral regression head introduced in `DSNT`_ by Nibali et al(2018). The head contains a differentiable spatial to numerical transform

Depthwise separable convolution module. See https://arxiv.org/pdf/1704.04861.pdf for details. This module can replace a ConvModule with the

Top-down integral regression head introduced in `IPR`_ by Xiao et al(2018). The head contains a differentiable spatial to numerical transform

The base data structure of MMPose that is used as the interface between modules. The attributes of ``PoseDataSample`` includes: - ``

A sequential module that passes timestep embeddings to the children that support it as an extra input.

Visualizer that draws data about detection/segmentation on images. It contains methods like `draw_{text,box,circle,line,binary_mask,polygon}

Computes and stores the average and current value

InvertedResidual block for MobileNetV2. Args: in_channels (int): The input channels of the InvertedResidual block. out_channels (

Inverted Residual Block. Args: in_channels (int): The input channels of this Module. out_channels (int): The output channels of t

Multi-stage multi-unit heatmap head introduced in `Multi-Stage Pose estimation Network (MSPN)`_ by Li et al (2019), and used by `Residual Steps

ModuleList in openmmlab. Args: modules (iterable, optional): an iterable of modules to add. init_cfg (dict, optional): Initializa

A residual block that can optionally change the number of channels. :param channels: the number of input channels. :param emb_channels: t

Top-down heatmap head introduced in `SimCC`_ by Li et al (2022). The head is composed of a few deconvolutional layers followed by a fully- con

A flexible Timer class. :Example: >>> import time >>> import annotator.uniformer.mmcv as mmcv >>> with mmcv.Timer(): >>> # s

Bottleneck block for ResNet. Args: in_channels (int): Input channels of this block. out_channels (int): Output channels of this b

NME evaluation metric. Calculate the normalized mean error (NME) of keypoints. Note: - length of dataset: N - num_keypoints:

Prepare sample for usage as network input.

Resize sample to given size (width, height).

SCNet backbone. Improving Convolutional Networks with Self-Calibrated Convolutions, Jiang-Jiang Liu, Qibin Hou, Ming-Ming Cheng, Changhu Wang

Get the bbox image as the model input by affine transform. Required Keys: - img - bbox_center - bbox_scale - bbo

Bottleneck block for ViPNAS_ResNet. Args: in_channels (int): Input channels of this block. out_channels (int): Output channels of

Basic block for VideoPose3D. Args: in_channels (int): Input channels of this block. out_channels (int): Output channels of this b

Encode keypoints into Target. The generated target is usually the supervision signal of the model learning, e.g. heatmaps or regression label

Normlize image by given mean and std.

Normlize image by given mean and std.

A residual 3D convolutional block. Args: in_channels (int): Input channels of this block. out_channels (int): Output channels of

Encode/decode keypoints with Structured Pose Representation (SPR). See the paper `Single-stage multi-person pose machines`_ by Nie et al (201

Swin Transformer A PyTorch implement of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows` - https://arxiv.

ViPNAS_MobileNetV3 backbone. "ViPNAS: Efficient Video Pose Estimation via Neural Architecture Search" More details can be found in the `paper

A PyTorch module for selecting, concatenating, and rescaling feature maps. Args: select_index (Optional[Union[int, Tuple[int]]], opti

High-Resolution Module for HRFormer. Args: num_branches (int): The number of branches in the HRFormerModule. block (nn.Module): T

Map a function over the elements in a dataset.

Normlize image by given mean and std.

The inferencer for 2D pose estimation. Args: model (str, optional): Pretrained 2D pose estimation algorithm. It's the path to

Implementation of `PVTv2: Improved Baselines with Pyramid Vision Transformer <https://arxiv.org/pdf/2106.13797.pdf>`_.

TCN backbone. Temporal Convolutional Networks. More details can be found in the `paper <https://arxiv.org/abs/1811.11742>`__ . Args:

VisPredictHead must be used together with other heads. It can predict keypoints coordinates of and their visibility simultaneously. In the cur

Encode/decode keypoints with the method introduced in "Associative Embedding". This is an asymmetric codec, where the keypoints are represente

BasicBlock for ResNet. Args: in_channels (int): Input channels of this block. out_channels (int): Output channels of this block.

Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks). We follow the implementation https://github.com

Pyramid Vision Transformer (PVT) Implementation of `Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolution

Top-down regression head introduced in `RLE`_ by Li et al(2021). The head is composed of fully-connected layers to predict the coordinates and

Region of interest feature map pooler that supports pooling from one or more feature maps.

Data augmentation with half-body transform that keeps only the upper or lower body at random. Required Keys: - keypoints - k

RegNet backbone. More details can be found in `paper <https://arxiv.org/abs/2003.13678>`__ . Args: arch (dict): The parameter of Reg

Resize the image to a target size.

Sequential module in openmmlab. Args: init_cfg (dict, optional): Initialization config dict.

ShuffleUnit block. ShuffleNet unit with pointwise group convolution (GConv) and channel shuffle. Args: in_channels (int): The in

Defines the Unet submodule with skip connection. X -------------------identity---------------------- |-- downsampling -- |submodule| -

ViPNAS_ResLayer to build ResNet style backbone. Args: block (nn.Module): Residual block used to build ViPNAS ResLayer. num_blocks

An attention block that allows spatial positions to attend to each other. Originally ported from here, but adapted to the N-d case. https

r"""Randomly shift, resize and rotate the image. Required Keys: - img - img_shape - keypoints (optional) Modified K

The box-to-box transform defined in R-CNN. The transformation is parameterized by 4 deltas: (dx, dy, dw, dh). The transformation scales the b

A wrapper of concatenated dataset. Same as :obj:`torch.utils.data.dataset.ConcatDataset`, but concat the group flag for image aspect ratio.

Encode/decode keypoints with the method introduced in the paper CID. See the paper Contextual Instance Decoupling for Robust Multi-Person Pos

Same as :class:`Checkpointer`, but is able to: 1. handle models in detectron & detectron2 model zoo, and apply conversions for legacy models.

Feature fusion block.

Feature fusion block.