Implements a module that performs conv2d + groupnorm + ReLU + Assumes kernel size is odd

This is a more standard version of the position embedding, very similar to the one used by the Attention is all you need paper, generalized t

Implements a module that performs conv2d + groupnorm + ReLU + conv2d + groupnorm + ReLU (and a possible downsam

SEGNET a Encoder-Decoder for Object Segmentation.

Implements a module that performs Upsample (reduction: conv2d + groupnorm + ReLU + bilinear_sampling) + concat + conv2d + gro

Very simple multi-layer perceptron (also called FFN)

This class computes an assignment between the targets and the predictions of the network For efficiency reasons, the targets don't include the no

r""" Modified from MultiheadAttention in PyTorch. Allows the model to jointly attend to information from different representation subspac

Make MeanShiftCrossAttentionLayer from CrossAttentionLayer.

Calculate the Munkres solution to the classical assignment problem. See the module documentation for usage.

Resize the image to a target size. Modified to support RGB-D image (W, H, 6)

This class computes the loss for DETR. The process happens in two steps: 1) we compute hungarian assignment between ground truth boxes and

Computes and stores the average and current value

Computes and stores the average and current value

A basic Swin Transformer layer for one stage. Args: dim (int): Number of feature channels depth (int): Depths of this stage.

Very simple multi-layer perceptron (also called FFN)

Very simple multi-layer perceptron (also called FFN)

Multilayer perceptron.

Image to Patch Embedding Args: patch_size (int): Patch token size. Default: 4. in_chans (int): Number of input image channels. Def

Resize image to a fixed target size

A SemanticSegmentor with test-time augmentation enabled. Its :meth:`__call__` method has the same interface as :meth:`SemanticSegmentor.forwa

Swin Transformer Block. Args: dim (int): Number of input channels. num_heads (int): Number of attention heads. window_size

Extension of the Trainer class adapted to MaskFormer.

A callable which takes a dataset dict in Detectron2 Dataset format, and map it into a format used by MaskFormer for instance segmentation.

Window based multi-head self attention (W-MSA) module with relative position bias. It supports both of shifted and non-shifted window. Args:

A callable which takes a dataset dict in Detectron2 Dataset format, and map it into a format used by MaskFormer. This dataset mapper app

A callable which takes a dataset dict in Detectron2 Dataset format, and map it into a format used by MaskFormer. This dataset mapper app

Evaluate AR for object proposals, AP for instance detection/segmentation, AP for keypoint detection outputs using COCO's metrics. See htt

A callable which takes a dataset dict in Detectron2 Dataset format, and map it into a format used by MaskFormer for instance segmentation.

A callable which takes a dataset dict in Detectron2 Dataset format, and map it into a format used by MaskFormer for panoptic segmentation.

A callable which takes a dataset dict in Detectron2 Dataset format, and map it into a format used by MaskFormer for semantic segmentation.

Main class for mask classification semantic segmentation architectures.

Patch Merging Layer Args: dim (int): Number of input channels. norm_layer (nn.Module, optional): Normalization layer. Default: nn

Main class for mask classification semantic segmentation architectures.

Instance-aware cropping.