Test segment without test-time aumengtation. Only the output of last decoder layers was used. Args: inputs (list[Tensor]

Placeholder for Forward function for training.

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

data format: B[spatial dims]C Returns pos_embedding: same format with data

Creates a downscale pyramid for the input tensor.

Generate proposals from encoded memory. Args: memory (Tensor) : The output of encoder, has shape (bs, num_key, em

Get the position embedding of proposal.

Get the reference points used in decoder. Args: spatial_shapes (Tensor): The shape of all feature maps, has shape

Get ``num_points`` most uncertain points with random points during train. Sample points in [0, 1] x [0, 1] coordinate space based on their

Estimate uncertainty based on pred logits. We estimate uncertainty as L1 distance between 0.0 and the logits prediction in 'mask_pred' for th

Get the valid radios of feature maps of all level.

Calculate lr decay rate for different ViT blocks. Args: name (string): parameter name. lr_decay_rate (float): base lr decay r

A context where the model is temporarily changed to eval mode, and restored to previous mode afterwards, Args: model: a torch Mo

Initialize layers of the DeformableDetrTransformer.

Returns: True if the current process is the main one.

Log no more than once per n seconds. Args: lvl (int): the logging level msg (str): n (int): name (str): name

Loss function. Args: all_cls_scores (Tensor): Classification scores for all decoder layers with shape [num_decode

Compute depth loss.

Compute depth loss.