The horizontal box class used in MMDetection by default. The ``box_dim`` of ``HorizontalBoxes`` is 4, which means the length of the last dime

A data structure interface of MMDetection. They are used as interfaces between different components. The attributes in ``DetDataSample`` are

This class represents masks in the form of bitmaps. Args: masks (ndarray): ndarray of masks in shape (N, H, W), where N is th

A basic convolution block including convolution, batch norm and ReLU. Args: in_channels (int): The number of input channels. out_

out_features = out_features or in_features\n hidden_features = hidden_features or in_features

Stores assignments between predicted and truth boxes. Attributes: num_gts (int): the number of truth boxes considered when computing this

Wrapper for linear function.

This class represents masks in the form of polygons. Polygons is a list of three levels. The first level of the list corresponds to objects,

ResNet backbone. Args: depth (int): Depth of resnet, from {18, 34, 50, 101, 152}. stem_channels (int | None): Number of stem chan

Image to Patch Embedding. We use a conv layer to implement PatchEmbed. Args: in_channels (int): The num of input channels. Default:

Rewrite Conv2d module according to DALL-E code.

A general data structure interface. It's used as the interface between different components. The following fields are convention names in MM

Random crop the image & bboxes & masks. The absolute ``crop_size`` is sampled based on ``crop_type`` and ``image_size``, then the cropped res

A data structure interface of tracking task in MMDetection. It is used as interfaces between different components. This data structure can be

Image pre-processor for detection tasks. Comparing with the :class:`mmengine.ImgDataPreprocessor`, 1. It supports batch augmentations. 2

Auto augmentation. This data augmentation is proposed in `AutoAugment: Learning Augmentation Policies from Data <https://arxiv.org/abs/1805.0

r"""Feature Pyramid Network. This is an implementation of paper `Feature Pyramid Networks for Object Detection <https://arxiv.org/abs/1612.03

COCO evaluation metric. Evaluate AR, AP, and mAP for detection tasks including proposal/box detection and instance segmentation. Please refer

Position encoding with sine and cosine functions. See `End-to-End Object Detection with Transformers <https://arxiv.org/pdf/2005.12872>`_ for

Object Detection Inferencer. Args: model (str, optional): Path to the config file or the model name defined in metafile. For

LVIS evaluation metric. Args: ann_file (str, optional): Path to the coco format annotation file. If not specified, ground tru

Standard points generator for multi-level (Mlvl) feature maps in 2D points-based detectors. Args: strides (list[int] | list[tuple[int

Applies padding to input (if needed) so that input can get fully covered by filter you specified. It support two modes "same" and "corner". The

COCO evaluation metric. Evaluate AR, AP, and mAP for detection tasks including proposal/box detection and instance segmentation. Please refer

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

Multi-head Attention Module. This module implements multi-head attention that supports different input dims and embed dims. And it also suppo

BottleBlock for TridentResNet. Args: trident_dilations (tuple[int, int, int]): Dilations of different trident branch.

CutOut operation. Randomly drop some regions of image used in `Cutout <https://arxiv.org/abs/1708.04552>`_. Required Keys: - img

Resize images & bbox & seg to the specified size. This transform resizes the input image according to ``width`` and ``height``. Bboxes, masks

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

Very simple multi-layer perceptron (also called FFN) with relu. Mostly used in DETR series detectors. Args: input_dim (int): Feature

A pseudo sampler that does not do sampling actually.

Random center crop and random around padding for CornerNet. This operation generates randomly cropped image from the original image and pads

This class is almost the same as official pycocotools package. It implements some snake case function aliases. So that the COCO class has the

Flip the image & bbox & mask & segmentation map. Added or Updated keys: flip, flip_direction, img, gt_bboxes, and gt_seg_map. There are 3 flip

A data structure interface of ReID task. It's used as interfaces between different components. Meta field: img_shape (Tuple): The sh

ResLayer to build ResNet style backbone for RPF in detectoRS. The difference between this module and base class is that we pass ``rfp_inplane

Resize images & bbox & seg. This transform resizes the input image according to ``scale`` or ``scale_factor``. Bboxes, masks, and seg map are

Shear the images, bboxes, masks and segmentation map vertically. Required Keys: - img - gt_bboxes (BaseBoxes[torch.float32]) (optional)

CSP-Darknet backbone used in YOLOv5 and YOLOX. Args: arch (str): Architecture of CSP-Darknet, from {P5, P6}. Default: P5.

An implementation of Conv2d + BatchNorm2d with support of fusion. Modified from https://github.com/microsoft/Cream/blob/main/TinyViT/models/t

Computes one-to-one matching between predictions and ground truth. This class computes an assignment between the targets and the predictions

LayerScale layer. Args: dim (int): Dimension of input features. layer_scale_init_value (float or torch.Tensor): Init value of lay

Rotate the images, bboxes, masks and segmentation map. Required Keys: - img - gt_bboxes (BaseBoxes[torch.float32]) (optional) - gt_m

Shear the images, bboxes, masks and segmentation map horizontally. Required Keys: - img - gt_bboxes (BaseBoxes[torch.float32]) (optional

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

Translate the images, bboxes, masks and segmentation map horizontally. Required Keys: - img - gt_bboxes (BaseBoxes[torch.float32]) (opti

Handle empty batch dimension to AdaptiveAvgPool2d.

COCO panoptic segmentation evaluation metric. Evaluate PQ, SQ RQ for panoptic segmentation tasks. Please refer to https://cocodataset.org/#pa

Merge patch feature map. This layer groups feature map by kernel_size, and applies norm and linear layers to the grouped feature map. Our imp

ResLayer to build ResNet style backbone. Args: block (nn.Module): block used to build ResLayer. inplanes (int): inplanes of block

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

Image pre-processor for tracking tasks. Accepts the data sampled by the dataloader, and preprocesses it into the format of the model

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

Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentation The simple copy-paste transform steps are as follows: 1. The

Base class for geometric transformations. All geometric transformations need to inherit from this base class. ``GeomTransform`` unifies the class

Assign a corresponding gt bbox or background to each bbox. Each proposals will be assigned with `-1`, `0`, or a positive integer indicating t

Load and process the ``instances`` and ``seg_map`` annotation provided by dataset. The annotation format is as the following: .. code-bl

Pad the image & segmentation map. There are three padding modes: (1) pad to a fixed size and (2) pad to the minimum size that is divisible by

Translate the images, bboxes, masks and segmentation map vertically. Required Keys: - img - gt_bboxes (BaseBoxes[torch.float32]) (option

Universal Visualizer for multiple tasks. Args: name (str): Name of the instance. Defaults to 'visualizer'. image (np.ndarray, opt

Assign a corresponding gt bbox or background to each prior. Each proposals will be assigned with `0` or a positive integer indicating the gro

Try to convert inputs to FP16 and CPU if got a PyTorch's CUDA Out of Memory error. It will do the following steps: 1. First retry after c

Base dataset for image classification task. This dataset support annotation file in `OpenMMLab 2.0 style annotation format`. .. _OpenMML

HiViT block. Args: input_size (int): Input size. dim (int): Number of input dims. num_heads (int): Number of attention he

Cross Stage Partial Layer. Args: in_channels (int): The input channels of the CSP layer. out_channels (int): The output channels

Assign pixels at the center region of a bbox as positive. Each proposals will be assigned with `-1`, `0`, or a positive integer indicating th

The dataloader benchmark class. It will be statistical inference FPS and CPU memory information. Args: cfg (mmengine.Config): config.

The dataset benchmark class. It will be statistical inference FPS, FPS pre transform and CPU memory information. Args: cfg (mmengine.

An implementation of LayerNorm of HorNet. The differences between HorNetLayerNorm & torch LayerNorm: 1. Supports two data formats channel

The inference benchmark class. It will be statistical inference FPS, CUDA memory and CPU memory information. Args: cfg (mmengine.Conf

Dataset for MOTChallenge. Args: visibility_thr (float, optional): The minimum visibility for the objects during training. Def

Assign a corresponding gt bbox or background to each bbox. Each proposals will be assigned with `-1`, or a semi-positive integer indicating t

Mosaic augmentation. Given 4 images, mosaic transform combines them into one output image. The output image is composed of the parts from eac

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

Random affine transform data augmentation. This operation randomly generates affine transform matrix which including rotation, translation, s

Extra layers of SSD backbone to generate multi-scale feature maps. Args: in_channels (Sequence[int]): Number of input channels per scale.

Simplified version of original basic residual block. This is used in `SCNet <https://arxiv.org/abs/2012.10150>`_. - Norm layer is now optiona

Computes 1-to-k matching between ground truth and predictions. This class computes an assignment between the targets and the predictions base

Encoder layer with window attention in Vision Transformer. Args: embed_dims (int): The feature dimension num_heads (int): Paralle

A sampler wrapper for grouping images with similar aspect ratio (< 1 or. >= 1) into a same batch. Args: sampler (Sampler): Base samp

Adjust the brightness of the image. A magnitude=0 gives a black image, whereas magnitude=1 gives the original image. The bboxes, masks and seg

Adjust the color balance of the image, in a manner similar to the controls on a colour TV set. A magnitude=0 gives a black & white image, wher

Control the contrast of the image. A magnitude=0 gives a gray image, whereas magnitude=1 gives the original imageThe bboxes, masks and segment

ResNet backbone for DetectoRS. Args: sac (dict, optional): Dictionary to construct SAC (Switchable Atrous Convolution). Defau

Randomly drop some regions of feature maps. Please refer to the method proposed in `DropBlock <https://arxiv.org/abs/1810.12890>`_ for deta

Dump model predictions to a pickle file for offline evaluation. Different from `DumpResults` in MMEngine, it compresses instance segmentation

Dynamic ReLU (DyReLU) module. See `Dynamic ReLU <https://arxiv.org/abs/2003.10027>`_ for details. Current implementation is specialized for t

Extract RoI features from all level feature maps levels. This is the implementation of `A novel Region of Interest Extraction Layer for Insta

HourglassNet backbone. Stacked Hourglass Networks for Human Pose Estimation. More details can be found in the `paper <https://arxiv.org/a