Parse the raw outputs (losses) of the network. Args: losses (dict): Raw output of the network, which usually contain

Transform inputs for decoder. Args: inputs (list[Tensor]): List of multi-level img features. Returns: Tensor

Calculate accuracy according to the prediction and target. Args: pred (torch.Tensor): The model prediction, shape (N, num_class, ...)

Build head.

Classify each pixel.

Encode images with backbone and decode into a semantic segmentation map of the same size as input.

Extract features from images.

Flattens predictions in the batch (binary case) Remove labels equal to 'ignore_index'.

Flattens predictions in the batch (binary case) Remove labels equal to 'ignore_index'.

Flattens predictions in the batch.

Flattens predictions in the batch.

Forward function.

Placeholder of forward function.

Placeholder of forward function.

generate 1d decay mask, the result is l*l

generate 1d depth decay mask, the result is l*l

Computes gradient of the Lovasz extension w.r.t sorted errors. See Alg. 1 in paper.

Computes gradient of the Lovasz extension w.r.t sorted errors. See Alg. 1 in paper.

Binary Lovasz hinge loss. Args: logits (torch.Tensor): [P], logits at each prediction (between -infty and +infty). la

Binary Lovasz hinge loss. Args: logits (torch.Tensor): [P], logits at each prediction (between -infty and +infty). la

Multi-class Lovasz-Softmax loss. Args: probs (torch.Tensor): [P, C], class probabilities at each prediction (between 0 and 1)

Multi-class Lovasz-Softmax loss. Args: probs (torch.Tensor): [P, C], class probabilities at each prediction (between 0 and 1)

nn.Module: normalization layer after the first convolution layer

nn.Module: normalization layer after the third convolution layer

Save ndarray to local numpy file. Args: array (ndarray): Ndarray to save. temp_file_name (str): Numpy file name. If 'temp_file_na

Resize pos_embed weights. Resize pos_embed using bicubic interpolate method. Args: pos_embed (torch.Tensor): Position emb

Resize relative pos_embed weights. This function is modified from https://github.com/microsoft/unilm/blob/master/beit/semantic_segmen

Sample pixels that have high loss or with low prediction confidence. Args: seg_logit (torch.Tensor): segmentation logits, shape (

Draw `result` over `img`. Args: img (str or Tensor): The image to be displayed. result (Tensor): The semantic segment

Calculate evaluation metrics Args: total_area_intersect (ndarray): The intersection of prediction and ground truth histogram o

Args: x: (B, H, W, C) Returns: windows: (num_windows*B, window_size, window_size, C)

Run forward function and calculate loss for auxiliary head in training.

Build transformer encoding layers.

Build patch embedding layer.

Check branches configuration.

Convert segmentation label to onehot. Args: seg_label (Tensor): Segmentation label of shape (N, H, W). num_classes (i

Run forward function and calculate loss for decode head in inference.

Run forward function and calculate loss for decode head in training.

Expand onehot labels to match the size of prediction.

Expand onehot labels to match the size of prediction.

Forward function for feature maps before classifying each pixel with ``self.cls_seg`` fc. Args: inputs (list[Tensor]): Li

Forward function for feature maps before classifying each pixel with ``self.cls_seg`` fc. Args: inputs (list[Tensor]): Li

Forward function for feature maps before classifying each pixel with ``self.cls_seg`` fc. Args: inputs (list[Tensor]): Li

Get new sequence via geometric sequence interpolation. Args: src_size (int): Pos_embedding size in pre-trained model.

Get model evaluation results.

Initialize ``auxiliary_head``

Initialize the W&B Tables for validation data.

Initialize ``decode_head``

Check and initialize input transforms. The in_channels, in_index and input_transform must match. Specifically, when input_transform i

Check and initialize input transforms. The in_channels, in_index and input_transform must match. Specifically, when input_transform i

Initialize the W&B Tables for model evaluation.