Load state dict, filtering out non-trainable modules if present.

Cumulative distribution function of the Gaussian distribution.

Remove non-trainable modules (LPIPS, tracker) from state dict before saving.

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

Expand the parameters to n_samples.

Inference with slide/whole style. Args: img (Tensor): The input image of shape (N, 3, H, W). img_meta (dict): Image i

Prediction each pixel.

Extract features from images.

Placeholder of forward function.

Args: imgs (List[Tensor]): the outer list indicates test-time augmentations and inner Tensor should have a shape

Forward function for training. Args: img (Tensor): Input images. img_metas (list[dict]): List of image info dict wher

this will perform the index select, cat the tensors, and provide the attn_bias from cache

this will perform the index select, cat the tensors, and provide the attn_bias from cache

Init. Args: features (int): number of features

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

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

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